diff --git a/cmd/zed/agents/zfs_agents.c b/cmd/zed/agents/zfs_agents.c index 2bc84a4f57d1..fb07266dae21 100644 --- a/cmd/zed/agents/zfs_agents.c +++ b/cmd/zed/agents/zfs_agents.c @@ -1,434 +1,458 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License Version 1.0 (CDDL-1.0). * You can obtain a copy of the license from the top-level file * "OPENSOLARIS.LICENSE" or at . * You may not use this file except in compliance with the license. * * CDDL HEADER END */ /* * Copyright (c) 2016, Intel Corporation. * Copyright (c) 2018, loli10K * Copyright (c) 2021 Hewlett Packard Enterprise Development LP */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_agents.h" #include "fmd_api.h" #include "../zed_log.h" /* * agent dispatch code */ static pthread_mutex_t agent_lock = PTHREAD_MUTEX_INITIALIZER; static pthread_cond_t agent_cond = PTHREAD_COND_INITIALIZER; static list_t agent_events; /* list of pending events */ static int agent_exiting; typedef struct agent_event { char ae_class[64]; char ae_subclass[32]; nvlist_t *ae_nvl; list_node_t ae_node; } agent_event_t; pthread_t g_agents_tid; libzfs_handle_t *g_zfs_hdl; /* guid search data */ typedef enum device_type { DEVICE_TYPE_L2ARC, /* l2arc device */ DEVICE_TYPE_SPARE, /* spare device */ DEVICE_TYPE_PRIMARY /* any primary pool storage device */ } device_type_t; typedef struct guid_search { uint64_t gs_pool_guid; uint64_t gs_vdev_guid; char *gs_devid; device_type_t gs_vdev_type; uint64_t gs_vdev_expandtime; /* vdev expansion time */ } guid_search_t; /* * Walks the vdev tree recursively looking for a matching devid. * Returns B_TRUE as soon as a matching device is found, B_FALSE otherwise. */ static boolean_t zfs_agent_iter_vdev(zpool_handle_t *zhp, nvlist_t *nvl, void *arg) { guid_search_t *gsp = arg; char *path = NULL; uint_t c, children; nvlist_t **child; + uint64_t vdev_guid; /* * First iterate over any children. */ if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { for (c = 0; c < children; c++) { if (zfs_agent_iter_vdev(zhp, child[c], gsp)) { gsp->gs_vdev_type = DEVICE_TYPE_PRIMARY; return (B_TRUE); } } } /* * Iterate over any spares and cache devices */ if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_SPARES, &child, &children) == 0) { for (c = 0; c < children; c++) { if (zfs_agent_iter_vdev(zhp, child[c], gsp)) { - gsp->gs_vdev_type = DEVICE_TYPE_L2ARC; + gsp->gs_vdev_type = DEVICE_TYPE_SPARE; return (B_TRUE); } } } if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_L2CACHE, &child, &children) == 0) { for (c = 0; c < children; c++) { if (zfs_agent_iter_vdev(zhp, child[c], gsp)) { - gsp->gs_vdev_type = DEVICE_TYPE_SPARE; + gsp->gs_vdev_type = DEVICE_TYPE_L2ARC; return (B_TRUE); } } } /* * On a devid match, grab the vdev guid and expansion time, if any. */ if (gsp->gs_devid != NULL && (nvlist_lookup_string(nvl, ZPOOL_CONFIG_DEVID, &path) == 0) && (strcmp(gsp->gs_devid, path) == 0)) { (void) nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_GUID, &gsp->gs_vdev_guid); (void) nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_EXPANSION_TIME, &gsp->gs_vdev_expandtime); return (B_TRUE); } + /* + * Otherwise, on a vdev guid match, grab the devid and expansion + * time. The devid might be missing on removal since its not part + * of blkid cache and L2ARC VDEV does not contain pool guid in its + * blkid, so this is a special case for L2ARC VDEV. + */ + else if (gsp->gs_vdev_guid != 0 && gsp->gs_devid == NULL && + nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_GUID, &vdev_guid) == 0 && + gsp->gs_vdev_guid == vdev_guid) { + (void) nvlist_lookup_string(nvl, ZPOOL_CONFIG_DEVID, + &gsp->gs_devid); + (void) nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_EXPANSION_TIME, + &gsp->gs_vdev_expandtime); + return (B_TRUE); + } return (B_FALSE); } static int zfs_agent_iter_pool(zpool_handle_t *zhp, void *arg) { guid_search_t *gsp = arg; nvlist_t *config, *nvl; /* * For each vdev in this pool, look for a match by devid */ if ((config = zpool_get_config(zhp, NULL)) != NULL) { if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) == 0) { (void) zfs_agent_iter_vdev(zhp, nvl, gsp); } } /* * if a match was found then grab the pool guid */ - if (gsp->gs_vdev_guid) { + if (gsp->gs_vdev_guid && gsp->gs_devid) { (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &gsp->gs_pool_guid); } zpool_close(zhp); return (gsp->gs_vdev_guid != 0); } void zfs_agent_post_event(const char *class, const char *subclass, nvlist_t *nvl) { agent_event_t *event; if (subclass == NULL) subclass = ""; event = malloc(sizeof (agent_event_t)); if (event == NULL || nvlist_dup(nvl, &event->ae_nvl, 0) != 0) { if (event) free(event); return; } if (strcmp(class, "sysevent.fs.zfs.vdev_check") == 0) { class = EC_ZFS; subclass = ESC_ZFS_VDEV_CHECK; } /* * On Linux, we don't get the expected FM_RESOURCE_REMOVED ereport * from the vdev_disk layer after a hot unplug. Fortunately we do * get an EC_DEV_REMOVE from our disk monitor and it is a suitable * proxy so we remap it here for the benefit of the diagnosis engine. * Starting in OpenZFS 2.0, we do get FM_RESOURCE_REMOVED from the spa * layer. Processing multiple FM_RESOURCE_REMOVED events is not harmful. */ if ((strcmp(class, EC_DEV_REMOVE) == 0) && (strcmp(subclass, ESC_DISK) == 0) && (nvlist_exists(nvl, ZFS_EV_VDEV_GUID) || nvlist_exists(nvl, DEV_IDENTIFIER))) { nvlist_t *payload = event->ae_nvl; struct timeval tv; int64_t tod[2]; uint64_t pool_guid = 0, vdev_guid = 0; guid_search_t search = { 0 }; device_type_t devtype = DEVICE_TYPE_PRIMARY; + char *devid = NULL; class = "resource.fs.zfs.removed"; subclass = ""; (void) nvlist_add_string(payload, FM_CLASS, class); + (void) nvlist_lookup_string(nvl, DEV_IDENTIFIER, &devid); (void) nvlist_lookup_uint64(nvl, ZFS_EV_POOL_GUID, &pool_guid); (void) nvlist_lookup_uint64(nvl, ZFS_EV_VDEV_GUID, &vdev_guid); (void) gettimeofday(&tv, NULL); tod[0] = tv.tv_sec; tod[1] = tv.tv_usec; (void) nvlist_add_int64_array(payload, FM_EREPORT_TIME, tod, 2); /* + * If devid is missing but vdev_guid is available, find devid + * and pool_guid from vdev_guid. * For multipath, spare and l2arc devices ZFS_EV_VDEV_GUID or * ZFS_EV_POOL_GUID may be missing so find them. */ - if (pool_guid == 0 || vdev_guid == 0) { - if ((nvlist_lookup_string(nvl, DEV_IDENTIFIER, - &search.gs_devid) == 0) && - (zpool_iter(g_zfs_hdl, zfs_agent_iter_pool, &search) - == 1)) { - if (pool_guid == 0) - pool_guid = search.gs_pool_guid; - if (vdev_guid == 0) - vdev_guid = search.gs_vdev_guid; - devtype = search.gs_vdev_type; - } + if (devid == NULL || pool_guid == 0 || vdev_guid == 0) { + if (devid == NULL) + search.gs_vdev_guid = vdev_guid; + else + search.gs_devid = devid; + zpool_iter(g_zfs_hdl, zfs_agent_iter_pool, &search); + if (devid == NULL) + devid = search.gs_devid; + if (pool_guid == 0) + pool_guid = search.gs_pool_guid; + if (vdev_guid == 0) + vdev_guid = search.gs_vdev_guid; + devtype = search.gs_vdev_type; } /* * We want to avoid reporting "remove" events coming from * libudev for VDEVs which were expanded recently (10s) and * avoid activating spares in response to partitions being * deleted and created in rapid succession. */ if (search.gs_vdev_expandtime != 0 && search.gs_vdev_expandtime + 10 > tv.tv_sec) { zed_log_msg(LOG_INFO, "agent post event: ignoring '%s' " "for recently expanded device '%s'", EC_DEV_REMOVE, - search.gs_devid); + devid); + fnvlist_free(payload); + free(event); goto out; } (void) nvlist_add_uint64(payload, FM_EREPORT_PAYLOAD_ZFS_POOL_GUID, pool_guid); (void) nvlist_add_uint64(payload, FM_EREPORT_PAYLOAD_ZFS_VDEV_GUID, vdev_guid); switch (devtype) { case DEVICE_TYPE_L2ARC: (void) nvlist_add_string(payload, FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE, VDEV_TYPE_L2CACHE); break; case DEVICE_TYPE_SPARE: (void) nvlist_add_string(payload, FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE, VDEV_TYPE_SPARE); break; case DEVICE_TYPE_PRIMARY: (void) nvlist_add_string(payload, FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE, VDEV_TYPE_DISK); break; } zed_log_msg(LOG_INFO, "agent post event: mapping '%s' to '%s'", EC_DEV_REMOVE, class); } (void) strlcpy(event->ae_class, class, sizeof (event->ae_class)); (void) strlcpy(event->ae_subclass, subclass, sizeof (event->ae_subclass)); (void) pthread_mutex_lock(&agent_lock); list_insert_tail(&agent_events, event); (void) pthread_mutex_unlock(&agent_lock); out: (void) pthread_cond_signal(&agent_cond); } static void zfs_agent_dispatch(const char *class, const char *subclass, nvlist_t *nvl) { /* * The diagnosis engine subscribes to the following events. * On illumos these subscriptions reside in: * /usr/lib/fm/fmd/plugins/zfs-diagnosis.conf */ if (strstr(class, "ereport.fs.zfs.") != NULL || strstr(class, "resource.fs.zfs.") != NULL || strcmp(class, "sysevent.fs.zfs.vdev_remove") == 0 || strcmp(class, "sysevent.fs.zfs.vdev_remove_dev") == 0 || strcmp(class, "sysevent.fs.zfs.pool_destroy") == 0) { fmd_module_recv(fmd_module_hdl("zfs-diagnosis"), nvl, class); } /* * The retire agent subscribes to the following events. * On illumos these subscriptions reside in: * /usr/lib/fm/fmd/plugins/zfs-retire.conf * * NOTE: faults events come directly from our diagnosis engine * and will not pass through the zfs kernel module. */ if (strcmp(class, FM_LIST_SUSPECT_CLASS) == 0 || strcmp(class, "resource.fs.zfs.removed") == 0 || strcmp(class, "resource.fs.zfs.statechange") == 0 || strcmp(class, "sysevent.fs.zfs.vdev_remove") == 0) { fmd_module_recv(fmd_module_hdl("zfs-retire"), nvl, class); } /* * The SLM module only consumes disk events and vdev check events * * NOTE: disk events come directly from disk monitor and will * not pass through the zfs kernel module. */ if (strstr(class, "EC_dev_") != NULL || strcmp(class, EC_ZFS) == 0) { (void) zfs_slm_event(class, subclass, nvl); } } /* * Events are consumed and dispatched from this thread * An agent can also post an event so event list lock * is not held when calling an agent. * One event is consumed at a time. */ static void * zfs_agent_consumer_thread(void *arg) { (void) arg; for (;;) { agent_event_t *event; (void) pthread_mutex_lock(&agent_lock); /* wait for an event to show up */ while (!agent_exiting && list_is_empty(&agent_events)) (void) pthread_cond_wait(&agent_cond, &agent_lock); if (agent_exiting) { (void) pthread_mutex_unlock(&agent_lock); zed_log_msg(LOG_INFO, "zfs_agent_consumer_thread: " "exiting"); return (NULL); } if ((event = (list_head(&agent_events))) != NULL) { list_remove(&agent_events, event); (void) pthread_mutex_unlock(&agent_lock); /* dispatch to all event subscribers */ zfs_agent_dispatch(event->ae_class, event->ae_subclass, event->ae_nvl); nvlist_free(event->ae_nvl); free(event); continue; } (void) pthread_mutex_unlock(&agent_lock); } return (NULL); } void zfs_agent_init(libzfs_handle_t *zfs_hdl) { fmd_hdl_t *hdl; g_zfs_hdl = zfs_hdl; if (zfs_slm_init() != 0) zed_log_die("Failed to initialize zfs slm"); zed_log_msg(LOG_INFO, "Add Agent: init"); hdl = fmd_module_hdl("zfs-diagnosis"); _zfs_diagnosis_init(hdl); if (!fmd_module_initialized(hdl)) zed_log_die("Failed to initialize zfs diagnosis"); hdl = fmd_module_hdl("zfs-retire"); _zfs_retire_init(hdl); if (!fmd_module_initialized(hdl)) zed_log_die("Failed to initialize zfs retire"); list_create(&agent_events, sizeof (agent_event_t), offsetof(struct agent_event, ae_node)); if (pthread_create(&g_agents_tid, NULL, zfs_agent_consumer_thread, NULL) != 0) { list_destroy(&agent_events); zed_log_die("Failed to initialize agents"); } pthread_setname_np(g_agents_tid, "agents"); } void zfs_agent_fini(void) { fmd_hdl_t *hdl; agent_event_t *event; agent_exiting = 1; (void) pthread_cond_signal(&agent_cond); /* wait for zfs_enum_pools thread to complete */ (void) pthread_join(g_agents_tid, NULL); /* drain any pending events */ while ((event = (list_head(&agent_events))) != NULL) { list_remove(&agent_events, event); nvlist_free(event->ae_nvl); free(event); } list_destroy(&agent_events); if ((hdl = fmd_module_hdl("zfs-retire")) != NULL) { _zfs_retire_fini(hdl); fmd_hdl_unregister(hdl); } if ((hdl = fmd_module_hdl("zfs-diagnosis")) != NULL) { _zfs_diagnosis_fini(hdl); fmd_hdl_unregister(hdl); } zed_log_msg(LOG_INFO, "Add Agent: fini"); zfs_slm_fini(); g_zfs_hdl = NULL; } diff --git a/cmd/zed/agents/zfs_retire.c b/cmd/zed/agents/zfs_retire.c index 5fe56fe81562..f4b6dff48176 100644 --- a/cmd/zed/agents/zfs_retire.c +++ b/cmd/zed/agents/zfs_retire.c @@ -1,566 +1,588 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved. * * Copyright (c) 2016, Intel Corporation. * Copyright (c) 2018, loli10K */ /* * The ZFS retire agent is responsible for managing hot spares across all pools. * When we see a device fault or a device removal, we try to open the associated * pool and look for any hot spares. We iterate over any available hot spares * and attempt a 'zpool replace' for each one. * * For vdevs diagnosed as faulty, the agent is also responsible for proactively * marking the vdev FAULTY (for I/O errors) or DEGRADED (for checksum errors). */ #include #include #include #include #include #include #include #include "zfs_agents.h" #include "fmd_api.h" typedef struct zfs_retire_repaired { struct zfs_retire_repaired *zrr_next; uint64_t zrr_pool; uint64_t zrr_vdev; } zfs_retire_repaired_t; typedef struct zfs_retire_data { libzfs_handle_t *zrd_hdl; zfs_retire_repaired_t *zrd_repaired; } zfs_retire_data_t; static void zfs_retire_clear_data(fmd_hdl_t *hdl, zfs_retire_data_t *zdp) { zfs_retire_repaired_t *zrp; while ((zrp = zdp->zrd_repaired) != NULL) { zdp->zrd_repaired = zrp->zrr_next; fmd_hdl_free(hdl, zrp, sizeof (zfs_retire_repaired_t)); } } /* * Find a pool with a matching GUID. */ typedef struct find_cbdata { uint64_t cb_guid; zpool_handle_t *cb_zhp; nvlist_t *cb_vdev; } find_cbdata_t; static int find_pool(zpool_handle_t *zhp, void *data) { find_cbdata_t *cbp = data; if (cbp->cb_guid == zpool_get_prop_int(zhp, ZPOOL_PROP_GUID, NULL)) { cbp->cb_zhp = zhp; return (1); } zpool_close(zhp); return (0); } /* * Find a vdev within a tree with a matching GUID. */ static nvlist_t * find_vdev(libzfs_handle_t *zhdl, nvlist_t *nv, uint64_t search_guid) { uint64_t guid; nvlist_t **child; uint_t c, children; nvlist_t *ret; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0 && guid == search_guid) { fmd_hdl_debug(fmd_module_hdl("zfs-retire"), "matched vdev %llu", guid); return (nv); } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (NULL); for (c = 0; c < children; c++) { if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL) return (ret); } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, &child, &children) != 0) return (NULL); for (c = 0; c < children; c++) { if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL) return (ret); } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, &child, &children) != 0) return (NULL); for (c = 0; c < children; c++) { if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL) return (ret); } return (NULL); } /* * Given a (pool, vdev) GUID pair, find the matching pool and vdev. */ static zpool_handle_t * find_by_guid(libzfs_handle_t *zhdl, uint64_t pool_guid, uint64_t vdev_guid, nvlist_t **vdevp) { find_cbdata_t cb; zpool_handle_t *zhp; nvlist_t *config, *nvroot; /* * Find the corresponding pool and make sure the vdev still exists. */ cb.cb_guid = pool_guid; if (zpool_iter(zhdl, find_pool, &cb) != 1) return (NULL); zhp = cb.cb_zhp; config = zpool_get_config(zhp, NULL); if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) != 0) { zpool_close(zhp); return (NULL); } if (vdev_guid != 0) { if ((*vdevp = find_vdev(zhdl, nvroot, vdev_guid)) == NULL) { zpool_close(zhp); return (NULL); } } return (zhp); } /* * Given a vdev, attempt to replace it with every known spare until one * succeeds or we run out of devices to try. * Return whether we were successful or not in replacing the device. */ static boolean_t replace_with_spare(fmd_hdl_t *hdl, zpool_handle_t *zhp, nvlist_t *vdev) { nvlist_t *config, *nvroot, *replacement; nvlist_t **spares; uint_t s, nspares; char *dev_name; zprop_source_t source; int ashift; config = zpool_get_config(zhp, NULL); if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) != 0) return (B_FALSE); /* * Find out if there are any hot spares available in the pool. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0) return (B_FALSE); /* * lookup "ashift" pool property, we may need it for the replacement */ ashift = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, &source); replacement = fmd_nvl_alloc(hdl, FMD_SLEEP); (void) nvlist_add_string(replacement, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT); dev_name = zpool_vdev_name(NULL, zhp, vdev, B_FALSE); /* * Try to replace each spare, ending when we successfully * replace it. */ for (s = 0; s < nspares; s++) { boolean_t rebuild = B_FALSE; char *spare_name, *type; if (nvlist_lookup_string(spares[s], ZPOOL_CONFIG_PATH, &spare_name) != 0) continue; /* prefer sequential resilvering for distributed spares */ if ((nvlist_lookup_string(spares[s], ZPOOL_CONFIG_TYPE, &type) == 0) && strcmp(type, VDEV_TYPE_DRAID_SPARE) == 0) rebuild = B_TRUE; /* if set, add the "ashift" pool property to the spare nvlist */ if (source != ZPROP_SRC_DEFAULT) (void) nvlist_add_uint64(spares[s], ZPOOL_CONFIG_ASHIFT, ashift); (void) nvlist_add_nvlist_array(replacement, ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)&spares[s], 1); fmd_hdl_debug(hdl, "zpool_vdev_replace '%s' with spare '%s'", dev_name, zfs_basename(spare_name)); if (zpool_vdev_attach(zhp, dev_name, spare_name, replacement, B_TRUE, rebuild) == 0) { free(dev_name); nvlist_free(replacement); return (B_TRUE); } } free(dev_name); nvlist_free(replacement); return (B_FALSE); } /* * Repair this vdev if we had diagnosed a 'fault.fs.zfs.device' and * ASRU is now usable. ZFS has found the device to be present and * functioning. */ static void zfs_vdev_repair(fmd_hdl_t *hdl, nvlist_t *nvl) { zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl); zfs_retire_repaired_t *zrp; uint64_t pool_guid, vdev_guid; if (nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_POOL_GUID, &pool_guid) != 0 || nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_GUID, &vdev_guid) != 0) return; /* * Before checking the state of the ASRU, go through and see if we've * already made an attempt to repair this ASRU. This list is cleared * whenever we receive any kind of list event, and is designed to * prevent us from generating a feedback loop when we attempt repairs * against a faulted pool. The problem is that checking the unusable * state of the ASRU can involve opening the pool, which can post * statechange events but otherwise leave the pool in the faulted * state. This list allows us to detect when a statechange event is * due to our own request. */ for (zrp = zdp->zrd_repaired; zrp != NULL; zrp = zrp->zrr_next) { if (zrp->zrr_pool == pool_guid && zrp->zrr_vdev == vdev_guid) return; } zrp = fmd_hdl_alloc(hdl, sizeof (zfs_retire_repaired_t), FMD_SLEEP); zrp->zrr_next = zdp->zrd_repaired; zrp->zrr_pool = pool_guid; zrp->zrr_vdev = vdev_guid; zdp->zrd_repaired = zrp; fmd_hdl_debug(hdl, "marking repaired vdev %llu on pool %llu", vdev_guid, pool_guid); } static void zfs_retire_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class) { (void) ep; uint64_t pool_guid, vdev_guid; zpool_handle_t *zhp; nvlist_t *resource, *fault; nvlist_t **faults; uint_t f, nfaults; zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl); libzfs_handle_t *zhdl = zdp->zrd_hdl; boolean_t fault_device, degrade_device; boolean_t is_repair; char *scheme; nvlist_t *vdev = NULL; char *uuid; int repair_done = 0; boolean_t retire; boolean_t is_disk; vdev_aux_t aux; uint64_t state = 0; + int l2arc; + vdev_stat_t *vs; + unsigned int c; fmd_hdl_debug(hdl, "zfs_retire_recv: '%s'", class); (void) nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_STATE, &state); /* * If this is a resource notifying us of device removal then simply * check for an available spare and continue unless the device is a * l2arc vdev, in which case we just offline it. */ if (strcmp(class, "resource.fs.zfs.removed") == 0 || (strcmp(class, "resource.fs.zfs.statechange") == 0 && (state == VDEV_STATE_REMOVED || state == VDEV_STATE_FAULTED))) { char *devtype; char *devname; if (nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_POOL_GUID, &pool_guid) != 0 || nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_GUID, &vdev_guid) != 0) return; if ((zhp = find_by_guid(zhdl, pool_guid, vdev_guid, &vdev)) == NULL) return; devname = zpool_vdev_name(NULL, zhp, vdev, B_FALSE); - /* Can't replace l2arc with a spare: offline the device */ - if (nvlist_lookup_string(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE, - &devtype) == 0 && strcmp(devtype, VDEV_TYPE_L2CACHE) == 0) { - fmd_hdl_debug(hdl, "zpool_vdev_offline '%s'", devname); - zpool_vdev_offline(zhp, devname, B_TRUE); - } else if (!fmd_prop_get_int32(hdl, "spare_on_remove") || - replace_with_spare(hdl, zhp, vdev) == B_FALSE) { + nvlist_lookup_uint64_array(vdev, ZPOOL_CONFIG_VDEV_STATS, + (uint64_t **)&vs, &c); + + /* + * If state removed is requested for already removed vdev, + * its a loopback event from spa_async_remove(). Just + * ignore it. + */ + if (vs->vs_state == VDEV_STATE_REMOVED && + state == VDEV_STATE_REMOVED) + return; + + l2arc = (nvlist_lookup_string(nvl, + FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE, &devtype) == 0 && + strcmp(devtype, VDEV_TYPE_L2CACHE) == 0); + + /* Remove the vdev since device is unplugged */ + if (l2arc || (strcmp(class, "resource.fs.zfs.removed") == 0)) { + int status = zpool_vdev_remove_wanted(zhp, devname); + fmd_hdl_debug(hdl, "zpool_vdev_remove_wanted '%s'" + ", ret:%d", devname, status); + } + + /* Replace the vdev with a spare if its not a l2arc */ + if (!l2arc && (!fmd_prop_get_int32(hdl, "spare_on_remove") || + replace_with_spare(hdl, zhp, vdev) == B_FALSE)) { /* Could not handle with spare */ fmd_hdl_debug(hdl, "no spare for '%s'", devname); } free(devname); zpool_close(zhp); return; } if (strcmp(class, FM_LIST_RESOLVED_CLASS) == 0) return; /* * Note: on Linux statechange events are more than just * healthy ones so we need to confirm the actual state value. */ if (strcmp(class, "resource.fs.zfs.statechange") == 0 && state == VDEV_STATE_HEALTHY) { zfs_vdev_repair(hdl, nvl); return; } if (strcmp(class, "sysevent.fs.zfs.vdev_remove") == 0) { zfs_vdev_repair(hdl, nvl); return; } zfs_retire_clear_data(hdl, zdp); if (strcmp(class, FM_LIST_REPAIRED_CLASS) == 0) is_repair = B_TRUE; else is_repair = B_FALSE; /* * We subscribe to zfs faults as well as all repair events. */ if (nvlist_lookup_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST, &faults, &nfaults) != 0) return; for (f = 0; f < nfaults; f++) { fault = faults[f]; fault_device = B_FALSE; degrade_device = B_FALSE; is_disk = B_FALSE; if (nvlist_lookup_boolean_value(fault, FM_SUSPECT_RETIRE, &retire) == 0 && retire == 0) continue; /* * While we subscribe to fault.fs.zfs.*, we only take action * for faults targeting a specific vdev (open failure or SERD * failure). We also subscribe to fault.io.* events, so that * faulty disks will be faulted in the ZFS configuration. */ if (fmd_nvl_class_match(hdl, fault, "fault.fs.zfs.vdev.io")) { fault_device = B_TRUE; } else if (fmd_nvl_class_match(hdl, fault, "fault.fs.zfs.vdev.checksum")) { degrade_device = B_TRUE; } else if (fmd_nvl_class_match(hdl, fault, "fault.fs.zfs.device")) { fault_device = B_FALSE; } else if (fmd_nvl_class_match(hdl, fault, "fault.io.*")) { is_disk = B_TRUE; fault_device = B_TRUE; } else { continue; } if (is_disk) { continue; } else { /* * This is a ZFS fault. Lookup the resource, and * attempt to find the matching vdev. */ if (nvlist_lookup_nvlist(fault, FM_FAULT_RESOURCE, &resource) != 0 || nvlist_lookup_string(resource, FM_FMRI_SCHEME, &scheme) != 0) continue; if (strcmp(scheme, FM_FMRI_SCHEME_ZFS) != 0) continue; if (nvlist_lookup_uint64(resource, FM_FMRI_ZFS_POOL, &pool_guid) != 0) continue; if (nvlist_lookup_uint64(resource, FM_FMRI_ZFS_VDEV, &vdev_guid) != 0) { if (is_repair) vdev_guid = 0; else continue; } if ((zhp = find_by_guid(zhdl, pool_guid, vdev_guid, &vdev)) == NULL) continue; aux = VDEV_AUX_ERR_EXCEEDED; } if (vdev_guid == 0) { /* * For pool-level repair events, clear the entire pool. */ fmd_hdl_debug(hdl, "zpool_clear of pool '%s'", zpool_get_name(zhp)); (void) zpool_clear(zhp, NULL, NULL); zpool_close(zhp); continue; } /* * If this is a repair event, then mark the vdev as repaired and * continue. */ if (is_repair) { repair_done = 1; fmd_hdl_debug(hdl, "zpool_clear of pool '%s' vdev %llu", zpool_get_name(zhp), vdev_guid); (void) zpool_vdev_clear(zhp, vdev_guid); zpool_close(zhp); continue; } /* * Actively fault the device if needed. */ if (fault_device) (void) zpool_vdev_fault(zhp, vdev_guid, aux); if (degrade_device) (void) zpool_vdev_degrade(zhp, vdev_guid, aux); if (fault_device || degrade_device) fmd_hdl_debug(hdl, "zpool_vdev_%s: vdev %llu on '%s'", fault_device ? "fault" : "degrade", vdev_guid, zpool_get_name(zhp)); /* * Attempt to substitute a hot spare. */ (void) replace_with_spare(hdl, zhp, vdev); zpool_close(zhp); } if (strcmp(class, FM_LIST_REPAIRED_CLASS) == 0 && repair_done && nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0) fmd_case_uuresolved(hdl, uuid); } static const fmd_hdl_ops_t fmd_ops = { zfs_retire_recv, /* fmdo_recv */ NULL, /* fmdo_timeout */ NULL, /* fmdo_close */ NULL, /* fmdo_stats */ NULL, /* fmdo_gc */ }; static const fmd_prop_t fmd_props[] = { { "spare_on_remove", FMD_TYPE_BOOL, "true" }, { NULL, 0, NULL } }; static const fmd_hdl_info_t fmd_info = { "ZFS Retire Agent", "1.0", &fmd_ops, fmd_props }; void _zfs_retire_init(fmd_hdl_t *hdl) { zfs_retire_data_t *zdp; libzfs_handle_t *zhdl; if ((zhdl = libzfs_init()) == NULL) return; if (fmd_hdl_register(hdl, FMD_API_VERSION, &fmd_info) != 0) { libzfs_fini(zhdl); return; } zdp = fmd_hdl_zalloc(hdl, sizeof (zfs_retire_data_t), FMD_SLEEP); zdp->zrd_hdl = zhdl; fmd_hdl_setspecific(hdl, zdp); } void _zfs_retire_fini(fmd_hdl_t *hdl) { zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl); if (zdp != NULL) { zfs_retire_clear_data(hdl, zdp); libzfs_fini(zdp->zrd_hdl); fmd_hdl_free(hdl, zdp, sizeof (zfs_retire_data_t)); } } diff --git a/cmd/ztest.c b/cmd/ztest.c index 436f5c156ec3..033bcfd33576 100644 --- a/cmd/ztest.c +++ b/cmd/ztest.c @@ -1,8289 +1,8289 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2018 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 Joyent, Inc. * Copyright (c) 2017, Intel Corporation. */ /* * The objective of this program is to provide a DMU/ZAP/SPA stress test * that runs entirely in userland, is easy to use, and easy to extend. * * The overall design of the ztest program is as follows: * * (1) For each major functional area (e.g. adding vdevs to a pool, * creating and destroying datasets, reading and writing objects, etc) * we have a simple routine to test that functionality. These * individual routines do not have to do anything "stressful". * * (2) We turn these simple functionality tests into a stress test by * running them all in parallel, with as many threads as desired, * and spread across as many datasets, objects, and vdevs as desired. * * (3) While all this is happening, we inject faults into the pool to * verify that self-healing data really works. * * (4) Every time we open a dataset, we change its checksum and compression * functions. Thus even individual objects vary from block to block * in which checksum they use and whether they're compressed. * * (5) To verify that we never lose on-disk consistency after a crash, * we run the entire test in a child of the main process. * At random times, the child self-immolates with a SIGKILL. * This is the software equivalent of pulling the power cord. * The parent then runs the test again, using the existing * storage pool, as many times as desired. If backwards compatibility * testing is enabled ztest will sometimes run the "older" version * of ztest after a SIGKILL. * * (6) To verify that we don't have future leaks or temporal incursions, * many of the functional tests record the transaction group number * as part of their data. When reading old data, they verify that * the transaction group number is less than the current, open txg. * If you add a new test, please do this if applicable. * * (7) Threads are created with a reduced stack size, for sanity checking. * Therefore, it's important not to allocate huge buffers on the stack. * * When run with no arguments, ztest runs for about five minutes and * produces no output if successful. To get a little bit of information, * specify -V. To get more information, specify -VV, and so on. * * To turn this into an overnight stress test, use -T to specify run time. * * You can ask more vdevs [-v], datasets [-d], or threads [-t] * to increase the pool capacity, fanout, and overall stress level. * * Use the -k option to set the desired frequency of kills. * * When ztest invokes itself it passes all relevant information through a * temporary file which is mmap-ed in the child process. This allows shared * memory to survive the exec syscall. The ztest_shared_hdr_t struct is always * stored at offset 0 of this file and contains information on the size and * number of shared structures in the file. The information stored in this file * must remain backwards compatible with older versions of ztest so that * ztest can invoke them during backwards compatibility testing (-B). */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if (__GLIBC__ && !__UCLIBC__) #include /* for backtrace() */ #endif static int ztest_fd_data = -1; static int ztest_fd_rand = -1; typedef struct ztest_shared_hdr { uint64_t zh_hdr_size; uint64_t zh_opts_size; uint64_t zh_size; uint64_t zh_stats_size; uint64_t zh_stats_count; uint64_t zh_ds_size; uint64_t zh_ds_count; } ztest_shared_hdr_t; static ztest_shared_hdr_t *ztest_shared_hdr; enum ztest_class_state { ZTEST_VDEV_CLASS_OFF, ZTEST_VDEV_CLASS_ON, ZTEST_VDEV_CLASS_RND }; #define ZO_GVARS_MAX_ARGLEN ((size_t)64) #define ZO_GVARS_MAX_COUNT ((size_t)10) typedef struct ztest_shared_opts { char zo_pool[ZFS_MAX_DATASET_NAME_LEN]; char zo_dir[ZFS_MAX_DATASET_NAME_LEN]; char zo_alt_ztest[MAXNAMELEN]; char zo_alt_libpath[MAXNAMELEN]; uint64_t zo_vdevs; uint64_t zo_vdevtime; size_t zo_vdev_size; int zo_ashift; int zo_mirrors; int zo_raid_children; int zo_raid_parity; char zo_raid_type[8]; int zo_draid_data; int zo_draid_spares; int zo_datasets; int zo_threads; uint64_t zo_passtime; uint64_t zo_killrate; int zo_verbose; int zo_init; uint64_t zo_time; uint64_t zo_maxloops; uint64_t zo_metaslab_force_ganging; int zo_mmp_test; int zo_special_vdevs; int zo_dump_dbgmsg; int zo_gvars_count; char zo_gvars[ZO_GVARS_MAX_COUNT][ZO_GVARS_MAX_ARGLEN]; } ztest_shared_opts_t; /* Default values for command line options. */ #define DEFAULT_POOL "ztest" #define DEFAULT_VDEV_DIR "/tmp" #define DEFAULT_VDEV_COUNT 5 #define DEFAULT_VDEV_SIZE (SPA_MINDEVSIZE * 4) /* 256m default size */ #define DEFAULT_VDEV_SIZE_STR "256M" #define DEFAULT_ASHIFT SPA_MINBLOCKSHIFT #define DEFAULT_MIRRORS 2 #define DEFAULT_RAID_CHILDREN 4 #define DEFAULT_RAID_PARITY 1 #define DEFAULT_DRAID_DATA 4 #define DEFAULT_DRAID_SPARES 1 #define DEFAULT_DATASETS_COUNT 7 #define DEFAULT_THREADS 23 #define DEFAULT_RUN_TIME 300 /* 300 seconds */ #define DEFAULT_RUN_TIME_STR "300 sec" #define DEFAULT_PASS_TIME 60 /* 60 seconds */ #define DEFAULT_PASS_TIME_STR "60 sec" #define DEFAULT_KILL_RATE 70 /* 70% kill rate */ #define DEFAULT_KILLRATE_STR "70%" #define DEFAULT_INITS 1 #define DEFAULT_MAX_LOOPS 50 /* 5 minutes */ #define DEFAULT_FORCE_GANGING (64 << 10) #define DEFAULT_FORCE_GANGING_STR "64K" /* Simplifying assumption: -1 is not a valid default. */ #define NO_DEFAULT -1 static const ztest_shared_opts_t ztest_opts_defaults = { .zo_pool = DEFAULT_POOL, .zo_dir = DEFAULT_VDEV_DIR, .zo_alt_ztest = { '\0' }, .zo_alt_libpath = { '\0' }, .zo_vdevs = DEFAULT_VDEV_COUNT, .zo_ashift = DEFAULT_ASHIFT, .zo_mirrors = DEFAULT_MIRRORS, .zo_raid_children = DEFAULT_RAID_CHILDREN, .zo_raid_parity = DEFAULT_RAID_PARITY, .zo_raid_type = VDEV_TYPE_RAIDZ, .zo_vdev_size = DEFAULT_VDEV_SIZE, .zo_draid_data = DEFAULT_DRAID_DATA, /* data drives */ .zo_draid_spares = DEFAULT_DRAID_SPARES, /* distributed spares */ .zo_datasets = DEFAULT_DATASETS_COUNT, .zo_threads = DEFAULT_THREADS, .zo_passtime = DEFAULT_PASS_TIME, .zo_killrate = DEFAULT_KILL_RATE, .zo_verbose = 0, .zo_mmp_test = 0, .zo_init = DEFAULT_INITS, .zo_time = DEFAULT_RUN_TIME, .zo_maxloops = DEFAULT_MAX_LOOPS, /* max loops during spa_freeze() */ .zo_metaslab_force_ganging = DEFAULT_FORCE_GANGING, .zo_special_vdevs = ZTEST_VDEV_CLASS_RND, .zo_gvars_count = 0, }; extern uint64_t metaslab_force_ganging; extern uint64_t metaslab_df_alloc_threshold; extern unsigned long zfs_deadman_synctime_ms; extern uint_t metaslab_preload_limit; extern int zfs_compressed_arc_enabled; extern int zfs_abd_scatter_enabled; extern uint_t dmu_object_alloc_chunk_shift; extern boolean_t zfs_force_some_double_word_sm_entries; extern unsigned long zio_decompress_fail_fraction; extern unsigned long zfs_reconstruct_indirect_damage_fraction; static ztest_shared_opts_t *ztest_shared_opts; static ztest_shared_opts_t ztest_opts; static const char *const ztest_wkeydata = "abcdefghijklmnopqrstuvwxyz012345"; typedef struct ztest_shared_ds { uint64_t zd_seq; } ztest_shared_ds_t; static ztest_shared_ds_t *ztest_shared_ds; #define ZTEST_GET_SHARED_DS(d) (&ztest_shared_ds[d]) #define BT_MAGIC 0x123456789abcdefULL #define MAXFAULTS(zs) \ (MAX((zs)->zs_mirrors, 1) * (ztest_opts.zo_raid_parity + 1) - 1) enum ztest_io_type { ZTEST_IO_WRITE_TAG, ZTEST_IO_WRITE_PATTERN, ZTEST_IO_WRITE_ZEROES, ZTEST_IO_TRUNCATE, ZTEST_IO_SETATTR, ZTEST_IO_REWRITE, ZTEST_IO_TYPES }; typedef struct ztest_block_tag { uint64_t bt_magic; uint64_t bt_objset; uint64_t bt_object; uint64_t bt_dnodesize; uint64_t bt_offset; uint64_t bt_gen; uint64_t bt_txg; uint64_t bt_crtxg; } ztest_block_tag_t; typedef struct bufwad { uint64_t bw_index; uint64_t bw_txg; uint64_t bw_data; } bufwad_t; /* * It would be better to use a rangelock_t per object. Unfortunately * the rangelock_t is not a drop-in replacement for rl_t, because we * still need to map from object ID to rangelock_t. */ typedef enum { RL_READER, RL_WRITER, RL_APPEND } rl_type_t; typedef struct rll { void *rll_writer; int rll_readers; kmutex_t rll_lock; kcondvar_t rll_cv; } rll_t; typedef struct rl { uint64_t rl_object; uint64_t rl_offset; uint64_t rl_size; rll_t *rl_lock; } rl_t; #define ZTEST_RANGE_LOCKS 64 #define ZTEST_OBJECT_LOCKS 64 /* * Object descriptor. Used as a template for object lookup/create/remove. */ typedef struct ztest_od { uint64_t od_dir; uint64_t od_object; dmu_object_type_t od_type; dmu_object_type_t od_crtype; uint64_t od_blocksize; uint64_t od_crblocksize; uint64_t od_crdnodesize; uint64_t od_gen; uint64_t od_crgen; char od_name[ZFS_MAX_DATASET_NAME_LEN]; } ztest_od_t; /* * Per-dataset state. */ typedef struct ztest_ds { ztest_shared_ds_t *zd_shared; objset_t *zd_os; pthread_rwlock_t zd_zilog_lock; zilog_t *zd_zilog; ztest_od_t *zd_od; /* debugging aid */ char zd_name[ZFS_MAX_DATASET_NAME_LEN]; kmutex_t zd_dirobj_lock; rll_t zd_object_lock[ZTEST_OBJECT_LOCKS]; rll_t zd_range_lock[ZTEST_RANGE_LOCKS]; } ztest_ds_t; /* * Per-iteration state. */ typedef void ztest_func_t(ztest_ds_t *zd, uint64_t id); typedef struct ztest_info { ztest_func_t *zi_func; /* test function */ uint64_t zi_iters; /* iterations per execution */ uint64_t *zi_interval; /* execute every seconds */ const char *zi_funcname; /* name of test function */ } ztest_info_t; typedef struct ztest_shared_callstate { uint64_t zc_count; /* per-pass count */ uint64_t zc_time; /* per-pass time */ uint64_t zc_next; /* next time to call this function */ } ztest_shared_callstate_t; static ztest_shared_callstate_t *ztest_shared_callstate; #define ZTEST_GET_SHARED_CALLSTATE(c) (&ztest_shared_callstate[c]) ztest_func_t ztest_dmu_read_write; ztest_func_t ztest_dmu_write_parallel; ztest_func_t ztest_dmu_object_alloc_free; ztest_func_t ztest_dmu_object_next_chunk; ztest_func_t ztest_dmu_commit_callbacks; ztest_func_t ztest_zap; ztest_func_t ztest_zap_parallel; ztest_func_t ztest_zil_commit; ztest_func_t ztest_zil_remount; ztest_func_t ztest_dmu_read_write_zcopy; ztest_func_t ztest_dmu_objset_create_destroy; ztest_func_t ztest_dmu_prealloc; ztest_func_t ztest_fzap; ztest_func_t ztest_dmu_snapshot_create_destroy; ztest_func_t ztest_dsl_prop_get_set; ztest_func_t ztest_spa_prop_get_set; ztest_func_t ztest_spa_create_destroy; ztest_func_t ztest_fault_inject; ztest_func_t ztest_dmu_snapshot_hold; ztest_func_t ztest_mmp_enable_disable; ztest_func_t ztest_scrub; ztest_func_t ztest_dsl_dataset_promote_busy; ztest_func_t ztest_vdev_attach_detach; ztest_func_t ztest_vdev_LUN_growth; ztest_func_t ztest_vdev_add_remove; ztest_func_t ztest_vdev_class_add; ztest_func_t ztest_vdev_aux_add_remove; ztest_func_t ztest_split_pool; ztest_func_t ztest_reguid; ztest_func_t ztest_spa_upgrade; ztest_func_t ztest_device_removal; ztest_func_t ztest_spa_checkpoint_create_discard; ztest_func_t ztest_initialize; ztest_func_t ztest_trim; ztest_func_t ztest_blake3; ztest_func_t ztest_fletcher; ztest_func_t ztest_fletcher_incr; ztest_func_t ztest_verify_dnode_bt; uint64_t zopt_always = 0ULL * NANOSEC; /* all the time */ uint64_t zopt_incessant = 1ULL * NANOSEC / 10; /* every 1/10 second */ uint64_t zopt_often = 1ULL * NANOSEC; /* every second */ uint64_t zopt_sometimes = 10ULL * NANOSEC; /* every 10 seconds */ uint64_t zopt_rarely = 60ULL * NANOSEC; /* every 60 seconds */ #define ZTI_INIT(func, iters, interval) \ { .zi_func = (func), \ .zi_iters = (iters), \ .zi_interval = (interval), \ .zi_funcname = # func } ztest_info_t ztest_info[] = { ZTI_INIT(ztest_dmu_read_write, 1, &zopt_always), ZTI_INIT(ztest_dmu_write_parallel, 10, &zopt_always), ZTI_INIT(ztest_dmu_object_alloc_free, 1, &zopt_always), ZTI_INIT(ztest_dmu_object_next_chunk, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_commit_callbacks, 1, &zopt_always), ZTI_INIT(ztest_zap, 30, &zopt_always), ZTI_INIT(ztest_zap_parallel, 100, &zopt_always), ZTI_INIT(ztest_split_pool, 1, &zopt_always), ZTI_INIT(ztest_zil_commit, 1, &zopt_incessant), ZTI_INIT(ztest_zil_remount, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_read_write_zcopy, 1, &zopt_often), ZTI_INIT(ztest_dmu_objset_create_destroy, 1, &zopt_often), ZTI_INIT(ztest_dsl_prop_get_set, 1, &zopt_often), ZTI_INIT(ztest_spa_prop_get_set, 1, &zopt_sometimes), #if 0 ZTI_INIT(ztest_dmu_prealloc, 1, &zopt_sometimes), #endif ZTI_INIT(ztest_fzap, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_snapshot_create_destroy, 1, &zopt_sometimes), ZTI_INIT(ztest_spa_create_destroy, 1, &zopt_sometimes), ZTI_INIT(ztest_fault_inject, 1, &zopt_sometimes), ZTI_INIT(ztest_dmu_snapshot_hold, 1, &zopt_sometimes), ZTI_INIT(ztest_mmp_enable_disable, 1, &zopt_sometimes), ZTI_INIT(ztest_reguid, 1, &zopt_rarely), ZTI_INIT(ztest_scrub, 1, &zopt_rarely), ZTI_INIT(ztest_spa_upgrade, 1, &zopt_rarely), ZTI_INIT(ztest_dsl_dataset_promote_busy, 1, &zopt_rarely), ZTI_INIT(ztest_vdev_attach_detach, 1, &zopt_sometimes), ZTI_INIT(ztest_vdev_LUN_growth, 1, &zopt_rarely), ZTI_INIT(ztest_vdev_add_remove, 1, &ztest_opts.zo_vdevtime), ZTI_INIT(ztest_vdev_class_add, 1, &ztest_opts.zo_vdevtime), ZTI_INIT(ztest_vdev_aux_add_remove, 1, &ztest_opts.zo_vdevtime), ZTI_INIT(ztest_device_removal, 1, &zopt_sometimes), ZTI_INIT(ztest_spa_checkpoint_create_discard, 1, &zopt_rarely), ZTI_INIT(ztest_initialize, 1, &zopt_sometimes), ZTI_INIT(ztest_trim, 1, &zopt_sometimes), ZTI_INIT(ztest_blake3, 1, &zopt_rarely), ZTI_INIT(ztest_fletcher, 1, &zopt_rarely), ZTI_INIT(ztest_fletcher_incr, 1, &zopt_rarely), ZTI_INIT(ztest_verify_dnode_bt, 1, &zopt_sometimes), }; #define ZTEST_FUNCS (sizeof (ztest_info) / sizeof (ztest_info_t)) /* * The following struct is used to hold a list of uncalled commit callbacks. * The callbacks are ordered by txg number. */ typedef struct ztest_cb_list { kmutex_t zcl_callbacks_lock; list_t zcl_callbacks; } ztest_cb_list_t; /* * Stuff we need to share writably between parent and child. */ typedef struct ztest_shared { boolean_t zs_do_init; hrtime_t zs_proc_start; hrtime_t zs_proc_stop; hrtime_t zs_thread_start; hrtime_t zs_thread_stop; hrtime_t zs_thread_kill; uint64_t zs_enospc_count; uint64_t zs_vdev_next_leaf; uint64_t zs_vdev_aux; uint64_t zs_alloc; uint64_t zs_space; uint64_t zs_splits; uint64_t zs_mirrors; uint64_t zs_metaslab_sz; uint64_t zs_metaslab_df_alloc_threshold; uint64_t zs_guid; } ztest_shared_t; #define ID_PARALLEL -1ULL static char ztest_dev_template[] = "%s/%s.%llua"; static char ztest_aux_template[] = "%s/%s.%s.%llu"; ztest_shared_t *ztest_shared; static spa_t *ztest_spa = NULL; static ztest_ds_t *ztest_ds; static kmutex_t ztest_vdev_lock; static boolean_t ztest_device_removal_active = B_FALSE; static boolean_t ztest_pool_scrubbed = B_FALSE; static kmutex_t ztest_checkpoint_lock; /* * The ztest_name_lock protects the pool and dataset namespace used by * the individual tests. To modify the namespace, consumers must grab * this lock as writer. Grabbing the lock as reader will ensure that the * namespace does not change while the lock is held. */ static pthread_rwlock_t ztest_name_lock; static boolean_t ztest_dump_core = B_TRUE; static boolean_t ztest_exiting; /* Global commit callback list */ static ztest_cb_list_t zcl; /* Commit cb delay */ static uint64_t zc_min_txg_delay = UINT64_MAX; static int zc_cb_counter = 0; /* * Minimum number of commit callbacks that need to be registered for us to check * whether the minimum txg delay is acceptable. */ #define ZTEST_COMMIT_CB_MIN_REG 100 /* * If a number of txgs equal to this threshold have been created after a commit * callback has been registered but not called, then we assume there is an * implementation bug. */ #define ZTEST_COMMIT_CB_THRESH (TXG_CONCURRENT_STATES + 1000) enum ztest_object { ZTEST_META_DNODE = 0, ZTEST_DIROBJ, ZTEST_OBJECTS }; static __attribute__((noreturn)) void usage(boolean_t requested); static int ztest_scrub_impl(spa_t *spa); /* * These libumem hooks provide a reasonable set of defaults for the allocator's * debugging facilities. */ const char * _umem_debug_init(void) { return ("default,verbose"); /* $UMEM_DEBUG setting */ } const char * _umem_logging_init(void) { return ("fail,contents"); /* $UMEM_LOGGING setting */ } static void dump_debug_buffer(void) { ssize_t ret __attribute__((unused)); if (!ztest_opts.zo_dump_dbgmsg) return; /* * We use write() instead of printf() so that this function * is safe to call from a signal handler. */ ret = write(STDOUT_FILENO, "\n", 1); zfs_dbgmsg_print("ztest"); } #define BACKTRACE_SZ 100 static void sig_handler(int signo) { struct sigaction action; #if (__GLIBC__ && !__UCLIBC__) /* backtrace() is a GNU extension */ int nptrs; void *buffer[BACKTRACE_SZ]; nptrs = backtrace(buffer, BACKTRACE_SZ); backtrace_symbols_fd(buffer, nptrs, STDERR_FILENO); #endif dump_debug_buffer(); /* * Restore default action and re-raise signal so SIGSEGV and * SIGABRT can trigger a core dump. */ action.sa_handler = SIG_DFL; sigemptyset(&action.sa_mask); action.sa_flags = 0; (void) sigaction(signo, &action, NULL); raise(signo); } #define FATAL_MSG_SZ 1024 static const char *fatal_msg; static __attribute__((format(printf, 2, 3))) __attribute__((noreturn)) void fatal(int do_perror, const char *message, ...) { va_list args; int save_errno = errno; char *buf; (void) fflush(stdout); buf = umem_alloc(FATAL_MSG_SZ, UMEM_NOFAIL); if (buf == NULL) goto out; va_start(args, message); (void) sprintf(buf, "ztest: "); /* LINTED */ (void) vsprintf(buf + strlen(buf), message, args); va_end(args); if (do_perror) { (void) snprintf(buf + strlen(buf), FATAL_MSG_SZ - strlen(buf), ": %s", strerror(save_errno)); } (void) fprintf(stderr, "%s\n", buf); fatal_msg = buf; /* to ease debugging */ out: if (ztest_dump_core) abort(); else dump_debug_buffer(); exit(3); } static int str2shift(const char *buf) { const char *ends = "BKMGTPEZ"; int i; if (buf[0] == '\0') return (0); for (i = 0; i < strlen(ends); i++) { if (toupper(buf[0]) == ends[i]) break; } if (i == strlen(ends)) { (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf); usage(B_FALSE); } if (buf[1] == '\0' || (toupper(buf[1]) == 'B' && buf[2] == '\0')) { return (10*i); } (void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf); usage(B_FALSE); } static uint64_t nicenumtoull(const char *buf) { char *end; uint64_t val; val = strtoull(buf, &end, 0); if (end == buf) { (void) fprintf(stderr, "ztest: bad numeric value: %s\n", buf); usage(B_FALSE); } else if (end[0] == '.') { double fval = strtod(buf, &end); fval *= pow(2, str2shift(end)); /* * UINT64_MAX is not exactly representable as a double. * The closest representation is UINT64_MAX + 1, so we * use a >= comparison instead of > for the bounds check. */ if (fval >= (double)UINT64_MAX) { (void) fprintf(stderr, "ztest: value too large: %s\n", buf); usage(B_FALSE); } val = (uint64_t)fval; } else { int shift = str2shift(end); if (shift >= 64 || (val << shift) >> shift != val) { (void) fprintf(stderr, "ztest: value too large: %s\n", buf); usage(B_FALSE); } val <<= shift; } return (val); } typedef struct ztest_option { const char short_opt; const char *long_opt; const char *long_opt_param; const char *comment; unsigned int default_int; const char *default_str; } ztest_option_t; /* * The following option_table is used for generating the usage info as well as * the long and short option information for calling getopt_long(). */ static ztest_option_t option_table[] = { { 'v', "vdevs", "INTEGER", "Number of vdevs", DEFAULT_VDEV_COUNT, NULL}, { 's', "vdev-size", "INTEGER", "Size of each vdev", NO_DEFAULT, DEFAULT_VDEV_SIZE_STR}, { 'a', "alignment-shift", "INTEGER", "Alignment shift; use 0 for random", DEFAULT_ASHIFT, NULL}, { 'm', "mirror-copies", "INTEGER", "Number of mirror copies", DEFAULT_MIRRORS, NULL}, { 'r', "raid-disks", "INTEGER", "Number of raidz/draid disks", DEFAULT_RAID_CHILDREN, NULL}, { 'R', "raid-parity", "INTEGER", "Raid parity", DEFAULT_RAID_PARITY, NULL}, { 'K', "raid-kind", "raidz|draid|random", "Raid kind", NO_DEFAULT, "random"}, { 'D', "draid-data", "INTEGER", "Number of draid data drives", DEFAULT_DRAID_DATA, NULL}, { 'S', "draid-spares", "INTEGER", "Number of draid spares", DEFAULT_DRAID_SPARES, NULL}, { 'd', "datasets", "INTEGER", "Number of datasets", DEFAULT_DATASETS_COUNT, NULL}, { 't', "threads", "INTEGER", "Number of ztest threads", DEFAULT_THREADS, NULL}, { 'g', "gang-block-threshold", "INTEGER", "Metaslab gang block threshold", NO_DEFAULT, DEFAULT_FORCE_GANGING_STR}, { 'i', "init-count", "INTEGER", "Number of times to initialize pool", DEFAULT_INITS, NULL}, { 'k', "kill-percentage", "INTEGER", "Kill percentage", NO_DEFAULT, DEFAULT_KILLRATE_STR}, { 'p', "pool-name", "STRING", "Pool name", NO_DEFAULT, DEFAULT_POOL}, { 'f', "vdev-file-directory", "PATH", "File directory for vdev files", NO_DEFAULT, DEFAULT_VDEV_DIR}, { 'M', "multi-host", NULL, "Multi-host; simulate pool imported on remote host", NO_DEFAULT, NULL}, { 'E', "use-existing-pool", NULL, "Use existing pool instead of creating new one", NO_DEFAULT, NULL}, { 'T', "run-time", "INTEGER", "Total run time", NO_DEFAULT, DEFAULT_RUN_TIME_STR}, { 'P', "pass-time", "INTEGER", "Time per pass", NO_DEFAULT, DEFAULT_PASS_TIME_STR}, { 'F', "freeze-loops", "INTEGER", "Max loops in spa_freeze()", DEFAULT_MAX_LOOPS, NULL}, { 'B', "alt-ztest", "PATH", "Alternate ztest path", NO_DEFAULT, NULL}, { 'C', "vdev-class-state", "on|off|random", "vdev class state", NO_DEFAULT, "random"}, { 'o', "option", "\"OPTION=INTEGER\"", "Set global variable to an unsigned 32-bit integer value", NO_DEFAULT, NULL}, { 'G', "dump-debug-msg", NULL, "Dump zfs_dbgmsg buffer before exiting due to an error", NO_DEFAULT, NULL}, { 'V', "verbose", NULL, "Verbose (use multiple times for ever more verbosity)", NO_DEFAULT, NULL}, { 'h', "help", NULL, "Show this help", NO_DEFAULT, NULL}, {0, 0, 0, 0, 0, 0} }; static struct option *long_opts = NULL; static char *short_opts = NULL; static void init_options(void) { ASSERT3P(long_opts, ==, NULL); ASSERT3P(short_opts, ==, NULL); int count = sizeof (option_table) / sizeof (option_table[0]); long_opts = umem_alloc(sizeof (struct option) * count, UMEM_NOFAIL); short_opts = umem_alloc(sizeof (char) * 2 * count, UMEM_NOFAIL); int short_opt_index = 0; for (int i = 0; i < count; i++) { long_opts[i].val = option_table[i].short_opt; long_opts[i].name = option_table[i].long_opt; long_opts[i].has_arg = option_table[i].long_opt_param != NULL ? required_argument : no_argument; long_opts[i].flag = NULL; short_opts[short_opt_index++] = option_table[i].short_opt; if (option_table[i].long_opt_param != NULL) { short_opts[short_opt_index++] = ':'; } } } static void fini_options(void) { int count = sizeof (option_table) / sizeof (option_table[0]); umem_free(long_opts, sizeof (struct option) * count); umem_free(short_opts, sizeof (char) * 2 * count); long_opts = NULL; short_opts = NULL; } static __attribute__((noreturn)) void usage(boolean_t requested) { char option[80]; FILE *fp = requested ? stdout : stderr; (void) fprintf(fp, "Usage: %s [OPTIONS...]\n", DEFAULT_POOL); for (int i = 0; option_table[i].short_opt != 0; i++) { if (option_table[i].long_opt_param != NULL) { (void) sprintf(option, " -%c --%s=%s", option_table[i].short_opt, option_table[i].long_opt, option_table[i].long_opt_param); } else { (void) sprintf(option, " -%c --%s", option_table[i].short_opt, option_table[i].long_opt); } (void) fprintf(fp, " %-40s%s", option, option_table[i].comment); if (option_table[i].long_opt_param != NULL) { if (option_table[i].default_str != NULL) { (void) fprintf(fp, " (default: %s)", option_table[i].default_str); } else if (option_table[i].default_int != NO_DEFAULT) { (void) fprintf(fp, " (default: %u)", option_table[i].default_int); } } (void) fprintf(fp, "\n"); } exit(requested ? 0 : 1); } static uint64_t ztest_random(uint64_t range) { uint64_t r; ASSERT3S(ztest_fd_rand, >=, 0); if (range == 0) return (0); if (read(ztest_fd_rand, &r, sizeof (r)) != sizeof (r)) fatal(B_TRUE, "short read from /dev/urandom"); return (r % range); } static void ztest_parse_name_value(const char *input, ztest_shared_opts_t *zo) { char name[32]; char *value; int state = ZTEST_VDEV_CLASS_RND; (void) strlcpy(name, input, sizeof (name)); value = strchr(name, '='); if (value == NULL) { (void) fprintf(stderr, "missing value in property=value " "'-C' argument (%s)\n", input); usage(B_FALSE); } *(value) = '\0'; value++; if (strcmp(value, "on") == 0) { state = ZTEST_VDEV_CLASS_ON; } else if (strcmp(value, "off") == 0) { state = ZTEST_VDEV_CLASS_OFF; } else if (strcmp(value, "random") == 0) { state = ZTEST_VDEV_CLASS_RND; } else { (void) fprintf(stderr, "invalid property value '%s'\n", value); usage(B_FALSE); } if (strcmp(name, "special") == 0) { zo->zo_special_vdevs = state; } else { (void) fprintf(stderr, "invalid property name '%s'\n", name); usage(B_FALSE); } if (zo->zo_verbose >= 3) (void) printf("%s vdev state is '%s'\n", name, value); } static void process_options(int argc, char **argv) { char *path; ztest_shared_opts_t *zo = &ztest_opts; int opt; uint64_t value; const char *raid_kind = "random"; memcpy(zo, &ztest_opts_defaults, sizeof (*zo)); init_options(); while ((opt = getopt_long(argc, argv, short_opts, long_opts, NULL)) != EOF) { value = 0; switch (opt) { case 'v': case 's': case 'a': case 'm': case 'r': case 'R': case 'D': case 'S': case 'd': case 't': case 'g': case 'i': case 'k': case 'T': case 'P': case 'F': value = nicenumtoull(optarg); } switch (opt) { case 'v': zo->zo_vdevs = value; break; case 's': zo->zo_vdev_size = MAX(SPA_MINDEVSIZE, value); break; case 'a': zo->zo_ashift = value; break; case 'm': zo->zo_mirrors = value; break; case 'r': zo->zo_raid_children = MAX(1, value); break; case 'R': zo->zo_raid_parity = MIN(MAX(value, 1), 3); break; case 'K': raid_kind = optarg; break; case 'D': zo->zo_draid_data = MAX(1, value); break; case 'S': zo->zo_draid_spares = MAX(1, value); break; case 'd': zo->zo_datasets = MAX(1, value); break; case 't': zo->zo_threads = MAX(1, value); break; case 'g': zo->zo_metaslab_force_ganging = MAX(SPA_MINBLOCKSIZE << 1, value); break; case 'i': zo->zo_init = value; break; case 'k': zo->zo_killrate = value; break; case 'p': (void) strlcpy(zo->zo_pool, optarg, sizeof (zo->zo_pool)); break; case 'f': path = realpath(optarg, NULL); if (path == NULL) { (void) fprintf(stderr, "error: %s: %s\n", optarg, strerror(errno)); usage(B_FALSE); } else { (void) strlcpy(zo->zo_dir, path, sizeof (zo->zo_dir)); free(path); } break; case 'M': zo->zo_mmp_test = 1; break; case 'V': zo->zo_verbose++; break; case 'E': zo->zo_init = 0; break; case 'T': zo->zo_time = value; break; case 'P': zo->zo_passtime = MAX(1, value); break; case 'F': zo->zo_maxloops = MAX(1, value); break; case 'B': (void) strlcpy(zo->zo_alt_ztest, optarg, sizeof (zo->zo_alt_ztest)); break; case 'C': ztest_parse_name_value(optarg, zo); break; case 'o': if (zo->zo_gvars_count >= ZO_GVARS_MAX_COUNT) { (void) fprintf(stderr, "max global var count (%zu) exceeded\n", ZO_GVARS_MAX_COUNT); usage(B_FALSE); } char *v = zo->zo_gvars[zo->zo_gvars_count]; if (strlcpy(v, optarg, ZO_GVARS_MAX_ARGLEN) >= ZO_GVARS_MAX_ARGLEN) { (void) fprintf(stderr, "global var option '%s' is too long\n", optarg); usage(B_FALSE); } zo->zo_gvars_count++; break; case 'G': zo->zo_dump_dbgmsg = 1; break; case 'h': usage(B_TRUE); break; case '?': default: usage(B_FALSE); break; } } fini_options(); /* When raid choice is 'random' add a draid pool 50% of the time */ if (strcmp(raid_kind, "random") == 0) { raid_kind = (ztest_random(2) == 0) ? "draid" : "raidz"; if (ztest_opts.zo_verbose >= 3) (void) printf("choosing RAID type '%s'\n", raid_kind); } if (strcmp(raid_kind, "draid") == 0) { uint64_t min_devsize; /* With fewer disk use 256M, otherwise 128M is OK */ min_devsize = (ztest_opts.zo_raid_children < 16) ? (256ULL << 20) : (128ULL << 20); /* No top-level mirrors with dRAID for now */ zo->zo_mirrors = 0; /* Use more appropriate defaults for dRAID */ if (zo->zo_vdevs == ztest_opts_defaults.zo_vdevs) zo->zo_vdevs = 1; if (zo->zo_raid_children == ztest_opts_defaults.zo_raid_children) zo->zo_raid_children = 16; if (zo->zo_ashift < 12) zo->zo_ashift = 12; if (zo->zo_vdev_size < min_devsize) zo->zo_vdev_size = min_devsize; if (zo->zo_draid_data + zo->zo_raid_parity > zo->zo_raid_children - zo->zo_draid_spares) { (void) fprintf(stderr, "error: too few draid " "children (%d) for stripe width (%d)\n", zo->zo_raid_children, zo->zo_draid_data + zo->zo_raid_parity); usage(B_FALSE); } (void) strlcpy(zo->zo_raid_type, VDEV_TYPE_DRAID, sizeof (zo->zo_raid_type)); } else /* using raidz */ { ASSERT0(strcmp(raid_kind, "raidz")); zo->zo_raid_parity = MIN(zo->zo_raid_parity, zo->zo_raid_children - 1); } zo->zo_vdevtime = (zo->zo_vdevs > 0 ? zo->zo_time * NANOSEC / zo->zo_vdevs : UINT64_MAX >> 2); if (*zo->zo_alt_ztest) { const char *invalid_what = "ztest"; char *val = zo->zo_alt_ztest; if (0 != access(val, X_OK) || (strrchr(val, '/') == NULL && (errno = EINVAL))) goto invalid; int dirlen = strrchr(val, '/') - val; strlcpy(zo->zo_alt_libpath, val, MIN(sizeof (zo->zo_alt_libpath), dirlen + 1)); invalid_what = "library path", val = zo->zo_alt_libpath; if (strrchr(val, '/') == NULL && (errno = EINVAL)) goto invalid; *strrchr(val, '/') = '\0'; strlcat(val, "/lib", sizeof (zo->zo_alt_libpath)); if (0 != access(zo->zo_alt_libpath, X_OK)) goto invalid; return; invalid: ztest_dump_core = B_FALSE; fatal(B_TRUE, "invalid alternate %s %s", invalid_what, val); } } static void ztest_kill(ztest_shared_t *zs) { zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(ztest_spa)); zs->zs_space = metaslab_class_get_space(spa_normal_class(ztest_spa)); /* * Before we kill ourselves, make sure that the config is updated. * See comment above spa_write_cachefile(). */ mutex_enter(&spa_namespace_lock); - spa_write_cachefile(ztest_spa, B_FALSE, B_FALSE); + spa_write_cachefile(ztest_spa, B_FALSE, B_FALSE, B_FALSE); mutex_exit(&spa_namespace_lock); (void) raise(SIGKILL); } static void ztest_record_enospc(const char *s) { (void) s; ztest_shared->zs_enospc_count++; } static uint64_t ztest_get_ashift(void) { if (ztest_opts.zo_ashift == 0) return (SPA_MINBLOCKSHIFT + ztest_random(5)); return (ztest_opts.zo_ashift); } static boolean_t ztest_is_draid_spare(const char *name) { uint64_t spare_id = 0, parity = 0, vdev_id = 0; if (sscanf(name, VDEV_TYPE_DRAID "%"PRIu64"-%"PRIu64"-%"PRIu64"", &parity, &vdev_id, &spare_id) == 3) { return (B_TRUE); } return (B_FALSE); } static nvlist_t * make_vdev_file(const char *path, const char *aux, const char *pool, size_t size, uint64_t ashift) { char *pathbuf = NULL; uint64_t vdev; nvlist_t *file; boolean_t draid_spare = B_FALSE; if (ashift == 0) ashift = ztest_get_ashift(); if (path == NULL) { pathbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); path = pathbuf; if (aux != NULL) { vdev = ztest_shared->zs_vdev_aux; (void) snprintf(pathbuf, MAXPATHLEN, ztest_aux_template, ztest_opts.zo_dir, pool == NULL ? ztest_opts.zo_pool : pool, aux, vdev); } else { vdev = ztest_shared->zs_vdev_next_leaf++; (void) snprintf(pathbuf, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, pool == NULL ? ztest_opts.zo_pool : pool, vdev); } } else { draid_spare = ztest_is_draid_spare(path); } if (size != 0 && !draid_spare) { int fd = open(path, O_RDWR | O_CREAT | O_TRUNC, 0666); if (fd == -1) fatal(B_TRUE, "can't open %s", path); if (ftruncate(fd, size) != 0) fatal(B_TRUE, "can't ftruncate %s", path); (void) close(fd); } file = fnvlist_alloc(); fnvlist_add_string(file, ZPOOL_CONFIG_TYPE, draid_spare ? VDEV_TYPE_DRAID_SPARE : VDEV_TYPE_FILE); fnvlist_add_string(file, ZPOOL_CONFIG_PATH, path); fnvlist_add_uint64(file, ZPOOL_CONFIG_ASHIFT, ashift); umem_free(pathbuf, MAXPATHLEN); return (file); } static nvlist_t * make_vdev_raid(const char *path, const char *aux, const char *pool, size_t size, uint64_t ashift, int r) { nvlist_t *raid, **child; int c; if (r < 2) return (make_vdev_file(path, aux, pool, size, ashift)); child = umem_alloc(r * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < r; c++) child[c] = make_vdev_file(path, aux, pool, size, ashift); raid = fnvlist_alloc(); fnvlist_add_string(raid, ZPOOL_CONFIG_TYPE, ztest_opts.zo_raid_type); fnvlist_add_uint64(raid, ZPOOL_CONFIG_NPARITY, ztest_opts.zo_raid_parity); fnvlist_add_nvlist_array(raid, ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)child, r); if (strcmp(ztest_opts.zo_raid_type, VDEV_TYPE_DRAID) == 0) { uint64_t ndata = ztest_opts.zo_draid_data; uint64_t nparity = ztest_opts.zo_raid_parity; uint64_t nspares = ztest_opts.zo_draid_spares; uint64_t children = ztest_opts.zo_raid_children; uint64_t ngroups = 1; /* * Calculate the minimum number of groups required to fill a * slice. This is the LCM of the stripe width (data + parity) * and the number of data drives (children - spares). */ while (ngroups * (ndata + nparity) % (children - nspares) != 0) ngroups++; /* Store the basic dRAID configuration. */ fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NDATA, ndata); fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NSPARES, nspares); fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NGROUPS, ngroups); } for (c = 0; c < r; c++) fnvlist_free(child[c]); umem_free(child, r * sizeof (nvlist_t *)); return (raid); } static nvlist_t * make_vdev_mirror(const char *path, const char *aux, const char *pool, size_t size, uint64_t ashift, int r, int m) { nvlist_t *mirror, **child; int c; if (m < 1) return (make_vdev_raid(path, aux, pool, size, ashift, r)); child = umem_alloc(m * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < m; c++) child[c] = make_vdev_raid(path, aux, pool, size, ashift, r); mirror = fnvlist_alloc(); fnvlist_add_string(mirror, ZPOOL_CONFIG_TYPE, VDEV_TYPE_MIRROR); fnvlist_add_nvlist_array(mirror, ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)child, m); for (c = 0; c < m; c++) fnvlist_free(child[c]); umem_free(child, m * sizeof (nvlist_t *)); return (mirror); } static nvlist_t * make_vdev_root(const char *path, const char *aux, const char *pool, size_t size, uint64_t ashift, const char *class, int r, int m, int t) { nvlist_t *root, **child; int c; boolean_t log; ASSERT3S(t, >, 0); log = (class != NULL && strcmp(class, "log") == 0); child = umem_alloc(t * sizeof (nvlist_t *), UMEM_NOFAIL); for (c = 0; c < t; c++) { child[c] = make_vdev_mirror(path, aux, pool, size, ashift, r, m); fnvlist_add_uint64(child[c], ZPOOL_CONFIG_IS_LOG, log); if (class != NULL && class[0] != '\0') { ASSERT(m > 1 || log); /* expecting a mirror */ fnvlist_add_string(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS, class); } } root = fnvlist_alloc(); fnvlist_add_string(root, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT); fnvlist_add_nvlist_array(root, aux ? aux : ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)child, t); for (c = 0; c < t; c++) fnvlist_free(child[c]); umem_free(child, t * sizeof (nvlist_t *)); return (root); } /* * Find a random spa version. Returns back a random spa version in the * range [initial_version, SPA_VERSION_FEATURES]. */ static uint64_t ztest_random_spa_version(uint64_t initial_version) { uint64_t version = initial_version; if (version <= SPA_VERSION_BEFORE_FEATURES) { version = version + ztest_random(SPA_VERSION_BEFORE_FEATURES - version + 1); } if (version > SPA_VERSION_BEFORE_FEATURES) version = SPA_VERSION_FEATURES; ASSERT(SPA_VERSION_IS_SUPPORTED(version)); return (version); } static int ztest_random_blocksize(void) { ASSERT3U(ztest_spa->spa_max_ashift, !=, 0); /* * Choose a block size >= the ashift. * If the SPA supports new MAXBLOCKSIZE, test up to 1MB blocks. */ int maxbs = SPA_OLD_MAXBLOCKSHIFT; if (spa_maxblocksize(ztest_spa) == SPA_MAXBLOCKSIZE) maxbs = 20; uint64_t block_shift = ztest_random(maxbs - ztest_spa->spa_max_ashift + 1); return (1 << (SPA_MINBLOCKSHIFT + block_shift)); } static int ztest_random_dnodesize(void) { int slots; int max_slots = spa_maxdnodesize(ztest_spa) >> DNODE_SHIFT; if (max_slots == DNODE_MIN_SLOTS) return (DNODE_MIN_SIZE); /* * Weight the random distribution more heavily toward smaller * dnode sizes since that is more likely to reflect real-world * usage. */ ASSERT3U(max_slots, >, 4); switch (ztest_random(10)) { case 0: slots = 5 + ztest_random(max_slots - 4); break; case 1 ... 4: slots = 2 + ztest_random(3); break; default: slots = 1; break; } return (slots << DNODE_SHIFT); } static int ztest_random_ibshift(void) { return (DN_MIN_INDBLKSHIFT + ztest_random(DN_MAX_INDBLKSHIFT - DN_MIN_INDBLKSHIFT + 1)); } static uint64_t ztest_random_vdev_top(spa_t *spa, boolean_t log_ok) { uint64_t top; vdev_t *rvd = spa->spa_root_vdev; vdev_t *tvd; ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0); do { top = ztest_random(rvd->vdev_children); tvd = rvd->vdev_child[top]; } while (!vdev_is_concrete(tvd) || (tvd->vdev_islog && !log_ok) || tvd->vdev_mg == NULL || tvd->vdev_mg->mg_class == NULL); return (top); } static uint64_t ztest_random_dsl_prop(zfs_prop_t prop) { uint64_t value; do { value = zfs_prop_random_value(prop, ztest_random(-1ULL)); } while (prop == ZFS_PROP_CHECKSUM && value == ZIO_CHECKSUM_OFF); return (value); } static int ztest_dsl_prop_set_uint64(char *osname, zfs_prop_t prop, uint64_t value, boolean_t inherit) { const char *propname = zfs_prop_to_name(prop); const char *valname; char *setpoint; uint64_t curval; int error; error = dsl_prop_set_int(osname, propname, (inherit ? ZPROP_SRC_NONE : ZPROP_SRC_LOCAL), value); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (error); } ASSERT0(error); setpoint = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); VERIFY0(dsl_prop_get_integer(osname, propname, &curval, setpoint)); if (ztest_opts.zo_verbose >= 6) { int err; err = zfs_prop_index_to_string(prop, curval, &valname); if (err) (void) printf("%s %s = %llu at '%s'\n", osname, propname, (unsigned long long)curval, setpoint); else (void) printf("%s %s = %s at '%s'\n", osname, propname, valname, setpoint); } umem_free(setpoint, MAXPATHLEN); return (error); } static int ztest_spa_prop_set_uint64(zpool_prop_t prop, uint64_t value) { spa_t *spa = ztest_spa; nvlist_t *props = NULL; int error; props = fnvlist_alloc(); fnvlist_add_uint64(props, zpool_prop_to_name(prop), value); error = spa_prop_set(spa, props); fnvlist_free(props); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (error); } ASSERT0(error); return (error); } static int ztest_dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, boolean_t decrypt, const void *tag, objset_t **osp) { int err; char *cp = NULL; char ddname[ZFS_MAX_DATASET_NAME_LEN]; strlcpy(ddname, name, sizeof (ddname)); cp = strchr(ddname, '@'); if (cp != NULL) *cp = '\0'; err = dmu_objset_own(name, type, readonly, decrypt, tag, osp); while (decrypt && err == EACCES) { dsl_crypto_params_t *dcp; nvlist_t *crypto_args = fnvlist_alloc(); fnvlist_add_uint8_array(crypto_args, "wkeydata", (uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN); VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, NULL, crypto_args, &dcp)); err = spa_keystore_load_wkey(ddname, dcp, B_FALSE); /* * Note: if there was an error loading, the wkey was not * consumed, and needs to be freed. */ dsl_crypto_params_free(dcp, (err != 0)); fnvlist_free(crypto_args); if (err == EINVAL) { /* * We couldn't load a key for this dataset so try * the parent. This loop will eventually hit the * encryption root since ztest only makes clones * as children of their origin datasets. */ cp = strrchr(ddname, '/'); if (cp == NULL) return (err); *cp = '\0'; err = EACCES; continue; } else if (err != 0) { break; } err = dmu_objset_own(name, type, readonly, decrypt, tag, osp); break; } return (err); } static void ztest_rll_init(rll_t *rll) { rll->rll_writer = NULL; rll->rll_readers = 0; mutex_init(&rll->rll_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&rll->rll_cv, NULL, CV_DEFAULT, NULL); } static void ztest_rll_destroy(rll_t *rll) { ASSERT3P(rll->rll_writer, ==, NULL); ASSERT0(rll->rll_readers); mutex_destroy(&rll->rll_lock); cv_destroy(&rll->rll_cv); } static void ztest_rll_lock(rll_t *rll, rl_type_t type) { mutex_enter(&rll->rll_lock); if (type == RL_READER) { while (rll->rll_writer != NULL) (void) cv_wait(&rll->rll_cv, &rll->rll_lock); rll->rll_readers++; } else { while (rll->rll_writer != NULL || rll->rll_readers) (void) cv_wait(&rll->rll_cv, &rll->rll_lock); rll->rll_writer = curthread; } mutex_exit(&rll->rll_lock); } static void ztest_rll_unlock(rll_t *rll) { mutex_enter(&rll->rll_lock); if (rll->rll_writer) { ASSERT0(rll->rll_readers); rll->rll_writer = NULL; } else { ASSERT3S(rll->rll_readers, >, 0); ASSERT3P(rll->rll_writer, ==, NULL); rll->rll_readers--; } if (rll->rll_writer == NULL && rll->rll_readers == 0) cv_broadcast(&rll->rll_cv); mutex_exit(&rll->rll_lock); } static void ztest_object_lock(ztest_ds_t *zd, uint64_t object, rl_type_t type) { rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)]; ztest_rll_lock(rll, type); } static void ztest_object_unlock(ztest_ds_t *zd, uint64_t object) { rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)]; ztest_rll_unlock(rll); } static rl_t * ztest_range_lock(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size, rl_type_t type) { uint64_t hash = object ^ (offset % (ZTEST_RANGE_LOCKS + 1)); rll_t *rll = &zd->zd_range_lock[hash & (ZTEST_RANGE_LOCKS - 1)]; rl_t *rl; rl = umem_alloc(sizeof (*rl), UMEM_NOFAIL); rl->rl_object = object; rl->rl_offset = offset; rl->rl_size = size; rl->rl_lock = rll; ztest_rll_lock(rll, type); return (rl); } static void ztest_range_unlock(rl_t *rl) { rll_t *rll = rl->rl_lock; ztest_rll_unlock(rll); umem_free(rl, sizeof (*rl)); } static void ztest_zd_init(ztest_ds_t *zd, ztest_shared_ds_t *szd, objset_t *os) { zd->zd_os = os; zd->zd_zilog = dmu_objset_zil(os); zd->zd_shared = szd; dmu_objset_name(os, zd->zd_name); int l; if (zd->zd_shared != NULL) zd->zd_shared->zd_seq = 0; VERIFY0(pthread_rwlock_init(&zd->zd_zilog_lock, NULL)); mutex_init(&zd->zd_dirobj_lock, NULL, MUTEX_DEFAULT, NULL); for (l = 0; l < ZTEST_OBJECT_LOCKS; l++) ztest_rll_init(&zd->zd_object_lock[l]); for (l = 0; l < ZTEST_RANGE_LOCKS; l++) ztest_rll_init(&zd->zd_range_lock[l]); } static void ztest_zd_fini(ztest_ds_t *zd) { int l; mutex_destroy(&zd->zd_dirobj_lock); (void) pthread_rwlock_destroy(&zd->zd_zilog_lock); for (l = 0; l < ZTEST_OBJECT_LOCKS; l++) ztest_rll_destroy(&zd->zd_object_lock[l]); for (l = 0; l < ZTEST_RANGE_LOCKS; l++) ztest_rll_destroy(&zd->zd_range_lock[l]); } #define TXG_MIGHTWAIT (ztest_random(10) == 0 ? TXG_NOWAIT : TXG_WAIT) static uint64_t ztest_tx_assign(dmu_tx_t *tx, uint64_t txg_how, const char *tag) { uint64_t txg; int error; /* * Attempt to assign tx to some transaction group. */ error = dmu_tx_assign(tx, txg_how); if (error) { if (error == ERESTART) { ASSERT3U(txg_how, ==, TXG_NOWAIT); dmu_tx_wait(tx); } else { ASSERT3U(error, ==, ENOSPC); ztest_record_enospc(tag); } dmu_tx_abort(tx); return (0); } txg = dmu_tx_get_txg(tx); ASSERT3U(txg, !=, 0); return (txg); } static void ztest_bt_generate(ztest_block_tag_t *bt, objset_t *os, uint64_t object, uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg, uint64_t crtxg) { bt->bt_magic = BT_MAGIC; bt->bt_objset = dmu_objset_id(os); bt->bt_object = object; bt->bt_dnodesize = dnodesize; bt->bt_offset = offset; bt->bt_gen = gen; bt->bt_txg = txg; bt->bt_crtxg = crtxg; } static void ztest_bt_verify(ztest_block_tag_t *bt, objset_t *os, uint64_t object, uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg, uint64_t crtxg) { ASSERT3U(bt->bt_magic, ==, BT_MAGIC); ASSERT3U(bt->bt_objset, ==, dmu_objset_id(os)); ASSERT3U(bt->bt_object, ==, object); ASSERT3U(bt->bt_dnodesize, ==, dnodesize); ASSERT3U(bt->bt_offset, ==, offset); ASSERT3U(bt->bt_gen, <=, gen); ASSERT3U(bt->bt_txg, <=, txg); ASSERT3U(bt->bt_crtxg, ==, crtxg); } static ztest_block_tag_t * ztest_bt_bonus(dmu_buf_t *db) { dmu_object_info_t doi; ztest_block_tag_t *bt; dmu_object_info_from_db(db, &doi); ASSERT3U(doi.doi_bonus_size, <=, db->db_size); ASSERT3U(doi.doi_bonus_size, >=, sizeof (*bt)); bt = (void *)((char *)db->db_data + doi.doi_bonus_size - sizeof (*bt)); return (bt); } /* * Generate a token to fill up unused bonus buffer space. Try to make * it unique to the object, generation, and offset to verify that data * is not getting overwritten by data from other dnodes. */ #define ZTEST_BONUS_FILL_TOKEN(obj, ds, gen, offset) \ (((ds) << 48) | ((gen) << 32) | ((obj) << 8) | (offset)) /* * Fill up the unused bonus buffer region before the block tag with a * verifiable pattern. Filling the whole bonus area with non-zero data * helps ensure that all dnode traversal code properly skips the * interior regions of large dnodes. */ static void ztest_fill_unused_bonus(dmu_buf_t *db, void *end, uint64_t obj, objset_t *os, uint64_t gen) { uint64_t *bonusp; ASSERT(IS_P2ALIGNED((char *)end - (char *)db->db_data, 8)); for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) { uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os), gen, bonusp - (uint64_t *)db->db_data); *bonusp = token; } } /* * Verify that the unused area of a bonus buffer is filled with the * expected tokens. */ static void ztest_verify_unused_bonus(dmu_buf_t *db, void *end, uint64_t obj, objset_t *os, uint64_t gen) { uint64_t *bonusp; for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) { uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os), gen, bonusp - (uint64_t *)db->db_data); VERIFY3U(*bonusp, ==, token); } } /* * ZIL logging ops */ #define lrz_type lr_mode #define lrz_blocksize lr_uid #define lrz_ibshift lr_gid #define lrz_bonustype lr_rdev #define lrz_dnodesize lr_crtime[1] static void ztest_log_create(ztest_ds_t *zd, dmu_tx_t *tx, lr_create_t *lr) { char *name = (void *)(lr + 1); /* name follows lr */ size_t namesize = strlen(name) + 1; itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_CREATE, sizeof (*lr) + namesize); memcpy(&itx->itx_lr + 1, &lr->lr_common + 1, sizeof (*lr) + namesize - sizeof (lr_t)); zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_remove(ztest_ds_t *zd, dmu_tx_t *tx, lr_remove_t *lr, uint64_t object) { char *name = (void *)(lr + 1); /* name follows lr */ size_t namesize = strlen(name) + 1; itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_REMOVE, sizeof (*lr) + namesize); memcpy(&itx->itx_lr + 1, &lr->lr_common + 1, sizeof (*lr) + namesize - sizeof (lr_t)); itx->itx_oid = object; zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_write(ztest_ds_t *zd, dmu_tx_t *tx, lr_write_t *lr) { itx_t *itx; itx_wr_state_t write_state = ztest_random(WR_NUM_STATES); if (zil_replaying(zd->zd_zilog, tx)) return; if (lr->lr_length > zil_max_log_data(zd->zd_zilog)) write_state = WR_INDIRECT; itx = zil_itx_create(TX_WRITE, sizeof (*lr) + (write_state == WR_COPIED ? lr->lr_length : 0)); if (write_state == WR_COPIED && dmu_read(zd->zd_os, lr->lr_foid, lr->lr_offset, lr->lr_length, ((lr_write_t *)&itx->itx_lr) + 1, DMU_READ_NO_PREFETCH) != 0) { zil_itx_destroy(itx); itx = zil_itx_create(TX_WRITE, sizeof (*lr)); write_state = WR_NEED_COPY; } itx->itx_private = zd; itx->itx_wr_state = write_state; itx->itx_sync = (ztest_random(8) == 0); memcpy(&itx->itx_lr + 1, &lr->lr_common + 1, sizeof (*lr) - sizeof (lr_t)); zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_truncate(ztest_ds_t *zd, dmu_tx_t *tx, lr_truncate_t *lr) { itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_TRUNCATE, sizeof (*lr)); memcpy(&itx->itx_lr + 1, &lr->lr_common + 1, sizeof (*lr) - sizeof (lr_t)); itx->itx_sync = B_FALSE; zil_itx_assign(zd->zd_zilog, itx, tx); } static void ztest_log_setattr(ztest_ds_t *zd, dmu_tx_t *tx, lr_setattr_t *lr) { itx_t *itx; if (zil_replaying(zd->zd_zilog, tx)) return; itx = zil_itx_create(TX_SETATTR, sizeof (*lr)); memcpy(&itx->itx_lr + 1, &lr->lr_common + 1, sizeof (*lr) - sizeof (lr_t)); itx->itx_sync = B_FALSE; zil_itx_assign(zd->zd_zilog, itx, tx); } /* * ZIL replay ops */ static int ztest_replay_create(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_create_t *lr = arg2; char *name = (void *)(lr + 1); /* name follows lr */ objset_t *os = zd->zd_os; ztest_block_tag_t *bbt; dmu_buf_t *db; dmu_tx_t *tx; uint64_t txg; int error = 0; int bonuslen; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ASSERT3U(lr->lr_doid, ==, ZTEST_DIROBJ); ASSERT3S(name[0], !=, '\0'); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, lr->lr_doid, B_TRUE, name); if (lr->lrz_type == DMU_OT_ZAP_OTHER) { dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, B_TRUE, NULL); } else { dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT); } txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) return (ENOSPC); ASSERT3U(dmu_objset_zil(os)->zl_replay, ==, !!lr->lr_foid); bonuslen = DN_BONUS_SIZE(lr->lrz_dnodesize); if (lr->lrz_type == DMU_OT_ZAP_OTHER) { if (lr->lr_foid == 0) { lr->lr_foid = zap_create_dnsize(os, lr->lrz_type, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } else { error = zap_create_claim_dnsize(os, lr->lr_foid, lr->lrz_type, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } } else { if (lr->lr_foid == 0) { lr->lr_foid = dmu_object_alloc_dnsize(os, lr->lrz_type, 0, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } else { error = dmu_object_claim_dnsize(os, lr->lr_foid, lr->lrz_type, 0, lr->lrz_bonustype, bonuslen, lr->lrz_dnodesize, tx); } } if (error) { ASSERT3U(error, ==, EEXIST); ASSERT(zd->zd_zilog->zl_replay); dmu_tx_commit(tx); return (error); } ASSERT3U(lr->lr_foid, !=, 0); if (lr->lrz_type != DMU_OT_ZAP_OTHER) VERIFY0(dmu_object_set_blocksize(os, lr->lr_foid, lr->lrz_blocksize, lr->lrz_ibshift, tx)); VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); bbt = ztest_bt_bonus(db); dmu_buf_will_dirty(db, tx); ztest_bt_generate(bbt, os, lr->lr_foid, lr->lrz_dnodesize, -1ULL, lr->lr_gen, txg, txg); ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, lr->lr_gen); dmu_buf_rele(db, FTAG); VERIFY0(zap_add(os, lr->lr_doid, name, sizeof (uint64_t), 1, &lr->lr_foid, tx)); (void) ztest_log_create(zd, tx, lr); dmu_tx_commit(tx); return (0); } static int ztest_replay_remove(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_remove_t *lr = arg2; char *name = (void *)(lr + 1); /* name follows lr */ objset_t *os = zd->zd_os; dmu_object_info_t doi; dmu_tx_t *tx; uint64_t object, txg; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ASSERT3U(lr->lr_doid, ==, ZTEST_DIROBJ); ASSERT3S(name[0], !=, '\0'); VERIFY0( zap_lookup(os, lr->lr_doid, name, sizeof (object), 1, &object)); ASSERT3U(object, !=, 0); ztest_object_lock(zd, object, RL_WRITER); VERIFY0(dmu_object_info(os, object, &doi)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, lr->lr_doid, B_FALSE, name); dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { ztest_object_unlock(zd, object); return (ENOSPC); } if (doi.doi_type == DMU_OT_ZAP_OTHER) { VERIFY0(zap_destroy(os, object, tx)); } else { VERIFY0(dmu_object_free(os, object, tx)); } VERIFY0(zap_remove(os, lr->lr_doid, name, tx)); (void) ztest_log_remove(zd, tx, lr, object); dmu_tx_commit(tx); ztest_object_unlock(zd, object); return (0); } static int ztest_replay_write(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_write_t *lr = arg2; objset_t *os = zd->zd_os; void *data = lr + 1; /* data follows lr */ uint64_t offset, length; ztest_block_tag_t *bt = data; ztest_block_tag_t *bbt; uint64_t gen, txg, lrtxg, crtxg; dmu_object_info_t doi; dmu_tx_t *tx; dmu_buf_t *db; arc_buf_t *abuf = NULL; rl_t *rl; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); offset = lr->lr_offset; length = lr->lr_length; /* If it's a dmu_sync() block, write the whole block */ if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) { uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr); if (length < blocksize) { offset -= offset % blocksize; length = blocksize; } } if (bt->bt_magic == BSWAP_64(BT_MAGIC)) byteswap_uint64_array(bt, sizeof (*bt)); if (bt->bt_magic != BT_MAGIC) bt = NULL; ztest_object_lock(zd, lr->lr_foid, RL_READER); rl = ztest_range_lock(zd, lr->lr_foid, offset, length, RL_WRITER); VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); dmu_object_info_from_db(db, &doi); bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); gen = bbt->bt_gen; crtxg = bbt->bt_crtxg; lrtxg = lr->lr_common.lrc_txg; tx = dmu_tx_create(os); dmu_tx_hold_write(tx, lr->lr_foid, offset, length); if (ztest_random(8) == 0 && length == doi.doi_data_block_size && P2PHASE(offset, length) == 0) abuf = dmu_request_arcbuf(db, length); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { if (abuf != NULL) dmu_return_arcbuf(abuf); dmu_buf_rele(db, FTAG); ztest_range_unlock(rl); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } if (bt != NULL) { /* * Usually, verify the old data before writing new data -- * but not always, because we also want to verify correct * behavior when the data was not recently read into cache. */ ASSERT0(offset % doi.doi_data_block_size); if (ztest_random(4) != 0) { int prefetch = ztest_random(2) ? DMU_READ_PREFETCH : DMU_READ_NO_PREFETCH; ztest_block_tag_t rbt; VERIFY(dmu_read(os, lr->lr_foid, offset, sizeof (rbt), &rbt, prefetch) == 0); if (rbt.bt_magic == BT_MAGIC) { ztest_bt_verify(&rbt, os, lr->lr_foid, 0, offset, gen, txg, crtxg); } } /* * Writes can appear to be newer than the bonus buffer because * the ztest_get_data() callback does a dmu_read() of the * open-context data, which may be different than the data * as it was when the write was generated. */ if (zd->zd_zilog->zl_replay) { ztest_bt_verify(bt, os, lr->lr_foid, 0, offset, MAX(gen, bt->bt_gen), MAX(txg, lrtxg), bt->bt_crtxg); } /* * Set the bt's gen/txg to the bonus buffer's gen/txg * so that all of the usual ASSERTs will work. */ ztest_bt_generate(bt, os, lr->lr_foid, 0, offset, gen, txg, crtxg); } if (abuf == NULL) { dmu_write(os, lr->lr_foid, offset, length, data, tx); } else { memcpy(abuf->b_data, data, length); VERIFY0(dmu_assign_arcbuf_by_dbuf(db, offset, abuf, tx)); } (void) ztest_log_write(zd, tx, lr); dmu_buf_rele(db, FTAG); dmu_tx_commit(tx); ztest_range_unlock(rl); ztest_object_unlock(zd, lr->lr_foid); return (0); } static int ztest_replay_truncate(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_truncate_t *lr = arg2; objset_t *os = zd->zd_os; dmu_tx_t *tx; uint64_t txg; rl_t *rl; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ztest_object_lock(zd, lr->lr_foid, RL_READER); rl = ztest_range_lock(zd, lr->lr_foid, lr->lr_offset, lr->lr_length, RL_WRITER); tx = dmu_tx_create(os); dmu_tx_hold_free(tx, lr->lr_foid, lr->lr_offset, lr->lr_length); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { ztest_range_unlock(rl); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } VERIFY0(dmu_free_range(os, lr->lr_foid, lr->lr_offset, lr->lr_length, tx)); (void) ztest_log_truncate(zd, tx, lr); dmu_tx_commit(tx); ztest_range_unlock(rl); ztest_object_unlock(zd, lr->lr_foid); return (0); } static int ztest_replay_setattr(void *arg1, void *arg2, boolean_t byteswap) { ztest_ds_t *zd = arg1; lr_setattr_t *lr = arg2; objset_t *os = zd->zd_os; dmu_tx_t *tx; dmu_buf_t *db; ztest_block_tag_t *bbt; uint64_t txg, lrtxg, crtxg, dnodesize; if (byteswap) byteswap_uint64_array(lr, sizeof (*lr)); ztest_object_lock(zd, lr->lr_foid, RL_WRITER); VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db)); tx = dmu_tx_create(os); dmu_tx_hold_bonus(tx, lr->lr_foid); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg == 0) { dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, lr->lr_foid); return (ENOSPC); } bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); crtxg = bbt->bt_crtxg; lrtxg = lr->lr_common.lrc_txg; dnodesize = bbt->bt_dnodesize; if (zd->zd_zilog->zl_replay) { ASSERT3U(lr->lr_size, !=, 0); ASSERT3U(lr->lr_mode, !=, 0); ASSERT3U(lrtxg, !=, 0); } else { /* * Randomly change the size and increment the generation. */ lr->lr_size = (ztest_random(db->db_size / sizeof (*bbt)) + 1) * sizeof (*bbt); lr->lr_mode = bbt->bt_gen + 1; ASSERT0(lrtxg); } /* * Verify that the current bonus buffer is not newer than our txg. */ ztest_bt_verify(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode, MAX(txg, lrtxg), crtxg); dmu_buf_will_dirty(db, tx); ASSERT3U(lr->lr_size, >=, sizeof (*bbt)); ASSERT3U(lr->lr_size, <=, db->db_size); VERIFY0(dmu_set_bonus(db, lr->lr_size, tx)); bbt = ztest_bt_bonus(db); ztest_bt_generate(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode, txg, crtxg); ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, bbt->bt_gen); dmu_buf_rele(db, FTAG); (void) ztest_log_setattr(zd, tx, lr); dmu_tx_commit(tx); ztest_object_unlock(zd, lr->lr_foid); return (0); } zil_replay_func_t *ztest_replay_vector[TX_MAX_TYPE] = { NULL, /* 0 no such transaction type */ ztest_replay_create, /* TX_CREATE */ NULL, /* TX_MKDIR */ NULL, /* TX_MKXATTR */ NULL, /* TX_SYMLINK */ ztest_replay_remove, /* TX_REMOVE */ NULL, /* TX_RMDIR */ NULL, /* TX_LINK */ NULL, /* TX_RENAME */ ztest_replay_write, /* TX_WRITE */ ztest_replay_truncate, /* TX_TRUNCATE */ ztest_replay_setattr, /* TX_SETATTR */ NULL, /* TX_ACL */ NULL, /* TX_CREATE_ACL */ NULL, /* TX_CREATE_ATTR */ NULL, /* TX_CREATE_ACL_ATTR */ NULL, /* TX_MKDIR_ACL */ NULL, /* TX_MKDIR_ATTR */ NULL, /* TX_MKDIR_ACL_ATTR */ NULL, /* TX_WRITE2 */ NULL, /* TX_SETSAXATTR */ }; /* * ZIL get_data callbacks */ static void ztest_get_done(zgd_t *zgd, int error) { (void) error; ztest_ds_t *zd = zgd->zgd_private; uint64_t object = ((rl_t *)zgd->zgd_lr)->rl_object; if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); ztest_range_unlock((rl_t *)zgd->zgd_lr); ztest_object_unlock(zd, object); umem_free(zgd, sizeof (*zgd)); } static int ztest_get_data(void *arg, uint64_t arg2, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio) { (void) arg2; ztest_ds_t *zd = arg; objset_t *os = zd->zd_os; uint64_t object = lr->lr_foid; uint64_t offset = lr->lr_offset; uint64_t size = lr->lr_length; uint64_t txg = lr->lr_common.lrc_txg; uint64_t crtxg; dmu_object_info_t doi; dmu_buf_t *db; zgd_t *zgd; int error; ASSERT3P(lwb, !=, NULL); ASSERT3P(zio, !=, NULL); ASSERT3U(size, !=, 0); ztest_object_lock(zd, object, RL_READER); error = dmu_bonus_hold(os, object, FTAG, &db); if (error) { ztest_object_unlock(zd, object); return (error); } crtxg = ztest_bt_bonus(db)->bt_crtxg; if (crtxg == 0 || crtxg > txg) { dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, object); return (ENOENT); } dmu_object_info_from_db(db, &doi); dmu_buf_rele(db, FTAG); db = NULL; zgd = umem_zalloc(sizeof (*zgd), UMEM_NOFAIL); zgd->zgd_lwb = lwb; zgd->zgd_private = zd; if (buf != NULL) { /* immediate write */ zgd->zgd_lr = (struct zfs_locked_range *)ztest_range_lock(zd, object, offset, size, RL_READER); error = dmu_read(os, object, offset, size, buf, DMU_READ_NO_PREFETCH); ASSERT0(error); } else { size = doi.doi_data_block_size; if (ISP2(size)) { offset = P2ALIGN(offset, size); } else { ASSERT3U(offset, <, size); offset = 0; } zgd->zgd_lr = (struct zfs_locked_range *)ztest_range_lock(zd, object, offset, size, RL_READER); error = dmu_buf_hold(os, object, offset, zgd, &db, DMU_READ_NO_PREFETCH); if (error == 0) { blkptr_t *bp = &lr->lr_blkptr; zgd->zgd_db = db; zgd->zgd_bp = bp; ASSERT3U(db->db_offset, ==, offset); ASSERT3U(db->db_size, ==, size); error = dmu_sync(zio, lr->lr_common.lrc_txg, ztest_get_done, zgd); if (error == 0) return (0); } } ztest_get_done(zgd, error); return (error); } static void * ztest_lr_alloc(size_t lrsize, char *name) { char *lr; size_t namesize = name ? strlen(name) + 1 : 0; lr = umem_zalloc(lrsize + namesize, UMEM_NOFAIL); if (name) memcpy(lr + lrsize, name, namesize); return (lr); } static void ztest_lr_free(void *lr, size_t lrsize, char *name) { size_t namesize = name ? strlen(name) + 1 : 0; umem_free(lr, lrsize + namesize); } /* * Lookup a bunch of objects. Returns the number of objects not found. */ static int ztest_lookup(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int error; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); for (i = 0; i < count; i++, od++) { od->od_object = 0; error = zap_lookup(zd->zd_os, od->od_dir, od->od_name, sizeof (uint64_t), 1, &od->od_object); if (error) { ASSERT3S(error, ==, ENOENT); ASSERT0(od->od_object); missing++; } else { dmu_buf_t *db; ztest_block_tag_t *bbt; dmu_object_info_t doi; ASSERT3U(od->od_object, !=, 0); ASSERT0(missing); /* there should be no gaps */ ztest_object_lock(zd, od->od_object, RL_READER); VERIFY0(dmu_bonus_hold(zd->zd_os, od->od_object, FTAG, &db)); dmu_object_info_from_db(db, &doi); bbt = ztest_bt_bonus(db); ASSERT3U(bbt->bt_magic, ==, BT_MAGIC); od->od_type = doi.doi_type; od->od_blocksize = doi.doi_data_block_size; od->od_gen = bbt->bt_gen; dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, od->od_object); } } return (missing); } static int ztest_create(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); for (i = 0; i < count; i++, od++) { if (missing) { od->od_object = 0; missing++; continue; } lr_create_t *lr = ztest_lr_alloc(sizeof (*lr), od->od_name); lr->lr_doid = od->od_dir; lr->lr_foid = 0; /* 0 to allocate, > 0 to claim */ lr->lrz_type = od->od_crtype; lr->lrz_blocksize = od->od_crblocksize; lr->lrz_ibshift = ztest_random_ibshift(); lr->lrz_bonustype = DMU_OT_UINT64_OTHER; lr->lrz_dnodesize = od->od_crdnodesize; lr->lr_gen = od->od_crgen; lr->lr_crtime[0] = time(NULL); if (ztest_replay_create(zd, lr, B_FALSE) != 0) { ASSERT0(missing); od->od_object = 0; missing++; } else { od->od_object = lr->lr_foid; od->od_type = od->od_crtype; od->od_blocksize = od->od_crblocksize; od->od_gen = od->od_crgen; ASSERT3U(od->od_object, !=, 0); } ztest_lr_free(lr, sizeof (*lr), od->od_name); } return (missing); } static int ztest_remove(ztest_ds_t *zd, ztest_od_t *od, int count) { int missing = 0; int error; int i; ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock)); od += count - 1; for (i = count - 1; i >= 0; i--, od--) { if (missing) { missing++; continue; } /* * No object was found. */ if (od->od_object == 0) continue; lr_remove_t *lr = ztest_lr_alloc(sizeof (*lr), od->od_name); lr->lr_doid = od->od_dir; if ((error = ztest_replay_remove(zd, lr, B_FALSE)) != 0) { ASSERT3U(error, ==, ENOSPC); missing++; } else { od->od_object = 0; } ztest_lr_free(lr, sizeof (*lr), od->od_name); } return (missing); } static int ztest_write(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size, void *data) { lr_write_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr) + size, NULL); lr->lr_foid = object; lr->lr_offset = offset; lr->lr_length = size; lr->lr_blkoff = 0; BP_ZERO(&lr->lr_blkptr); memcpy(lr + 1, data, size); error = ztest_replay_write(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr) + size, NULL); return (error); } static int ztest_truncate(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size) { lr_truncate_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr), NULL); lr->lr_foid = object; lr->lr_offset = offset; lr->lr_length = size; error = ztest_replay_truncate(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr), NULL); return (error); } static int ztest_setattr(ztest_ds_t *zd, uint64_t object) { lr_setattr_t *lr; int error; lr = ztest_lr_alloc(sizeof (*lr), NULL); lr->lr_foid = object; lr->lr_size = 0; lr->lr_mode = 0; error = ztest_replay_setattr(zd, lr, B_FALSE); ztest_lr_free(lr, sizeof (*lr), NULL); return (error); } static void ztest_prealloc(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size) { objset_t *os = zd->zd_os; dmu_tx_t *tx; uint64_t txg; rl_t *rl; txg_wait_synced(dmu_objset_pool(os), 0); ztest_object_lock(zd, object, RL_READER); rl = ztest_range_lock(zd, object, offset, size, RL_WRITER); tx = dmu_tx_create(os); dmu_tx_hold_write(tx, object, offset, size); txg = ztest_tx_assign(tx, TXG_WAIT, FTAG); if (txg != 0) { dmu_prealloc(os, object, offset, size, tx); dmu_tx_commit(tx); txg_wait_synced(dmu_objset_pool(os), txg); } else { (void) dmu_free_long_range(os, object, offset, size); } ztest_range_unlock(rl); ztest_object_unlock(zd, object); } static void ztest_io(ztest_ds_t *zd, uint64_t object, uint64_t offset) { int err; ztest_block_tag_t wbt; dmu_object_info_t doi; enum ztest_io_type io_type; uint64_t blocksize; void *data; VERIFY0(dmu_object_info(zd->zd_os, object, &doi)); blocksize = doi.doi_data_block_size; data = umem_alloc(blocksize, UMEM_NOFAIL); /* * Pick an i/o type at random, biased toward writing block tags. */ io_type = ztest_random(ZTEST_IO_TYPES); if (ztest_random(2) == 0) io_type = ZTEST_IO_WRITE_TAG; (void) pthread_rwlock_rdlock(&zd->zd_zilog_lock); switch (io_type) { case ZTEST_IO_WRITE_TAG: ztest_bt_generate(&wbt, zd->zd_os, object, doi.doi_dnodesize, offset, 0, 0, 0); (void) ztest_write(zd, object, offset, sizeof (wbt), &wbt); break; case ZTEST_IO_WRITE_PATTERN: (void) memset(data, 'a' + (object + offset) % 5, blocksize); if (ztest_random(2) == 0) { /* * Induce fletcher2 collisions to ensure that * zio_ddt_collision() detects and resolves them * when using fletcher2-verify for deduplication. */ ((uint64_t *)data)[0] ^= 1ULL << 63; ((uint64_t *)data)[4] ^= 1ULL << 63; } (void) ztest_write(zd, object, offset, blocksize, data); break; case ZTEST_IO_WRITE_ZEROES: memset(data, 0, blocksize); (void) ztest_write(zd, object, offset, blocksize, data); break; case ZTEST_IO_TRUNCATE: (void) ztest_truncate(zd, object, offset, blocksize); break; case ZTEST_IO_SETATTR: (void) ztest_setattr(zd, object); break; default: break; case ZTEST_IO_REWRITE: (void) pthread_rwlock_rdlock(&ztest_name_lock); err = ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_CHECKSUM, spa_dedup_checksum(ztest_spa), B_FALSE); VERIFY(err == 0 || err == ENOSPC); err = ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_COMPRESSION, ztest_random_dsl_prop(ZFS_PROP_COMPRESSION), B_FALSE); VERIFY(err == 0 || err == ENOSPC); (void) pthread_rwlock_unlock(&ztest_name_lock); VERIFY0(dmu_read(zd->zd_os, object, offset, blocksize, data, DMU_READ_NO_PREFETCH)); (void) ztest_write(zd, object, offset, blocksize, data); break; } (void) pthread_rwlock_unlock(&zd->zd_zilog_lock); umem_free(data, blocksize); } /* * Initialize an object description template. */ static void ztest_od_init(ztest_od_t *od, uint64_t id, const char *tag, uint64_t index, dmu_object_type_t type, uint64_t blocksize, uint64_t dnodesize, uint64_t gen) { od->od_dir = ZTEST_DIROBJ; od->od_object = 0; od->od_crtype = type; od->od_crblocksize = blocksize ? blocksize : ztest_random_blocksize(); od->od_crdnodesize = dnodesize ? dnodesize : ztest_random_dnodesize(); od->od_crgen = gen; od->od_type = DMU_OT_NONE; od->od_blocksize = 0; od->od_gen = 0; (void) snprintf(od->od_name, sizeof (od->od_name), "%s(%"PRId64")[%"PRIu64"]", tag, id, index); } /* * Lookup or create the objects for a test using the od template. * If the objects do not all exist, or if 'remove' is specified, * remove any existing objects and create new ones. Otherwise, * use the existing objects. */ static int ztest_object_init(ztest_ds_t *zd, ztest_od_t *od, size_t size, boolean_t remove) { int count = size / sizeof (*od); int rv = 0; mutex_enter(&zd->zd_dirobj_lock); if ((ztest_lookup(zd, od, count) != 0 || remove) && (ztest_remove(zd, od, count) != 0 || ztest_create(zd, od, count) != 0)) rv = -1; zd->zd_od = od; mutex_exit(&zd->zd_dirobj_lock); return (rv); } void ztest_zil_commit(ztest_ds_t *zd, uint64_t id) { (void) id; zilog_t *zilog = zd->zd_zilog; (void) pthread_rwlock_rdlock(&zd->zd_zilog_lock); zil_commit(zilog, ztest_random(ZTEST_OBJECTS)); /* * Remember the committed values in zd, which is in parent/child * shared memory. If we die, the next iteration of ztest_run() * will verify that the log really does contain this record. */ mutex_enter(&zilog->zl_lock); ASSERT3P(zd->zd_shared, !=, NULL); ASSERT3U(zd->zd_shared->zd_seq, <=, zilog->zl_commit_lr_seq); zd->zd_shared->zd_seq = zilog->zl_commit_lr_seq; mutex_exit(&zilog->zl_lock); (void) pthread_rwlock_unlock(&zd->zd_zilog_lock); } /* * This function is designed to simulate the operations that occur during a * mount/unmount operation. We hold the dataset across these operations in an * attempt to expose any implicit assumptions about ZIL management. */ void ztest_zil_remount(ztest_ds_t *zd, uint64_t id) { (void) id; objset_t *os = zd->zd_os; /* * We hold the ztest_vdev_lock so we don't cause problems with * other threads that wish to remove a log device, such as * ztest_device_removal(). */ mutex_enter(&ztest_vdev_lock); /* * We grab the zd_dirobj_lock to ensure that no other thread is * updating the zil (i.e. adding in-memory log records) and the * zd_zilog_lock to block any I/O. */ mutex_enter(&zd->zd_dirobj_lock); (void) pthread_rwlock_wrlock(&zd->zd_zilog_lock); /* zfsvfs_teardown() */ zil_close(zd->zd_zilog); /* zfsvfs_setup() */ VERIFY3P(zil_open(os, ztest_get_data, NULL), ==, zd->zd_zilog); zil_replay(os, zd, ztest_replay_vector); (void) pthread_rwlock_unlock(&zd->zd_zilog_lock); mutex_exit(&zd->zd_dirobj_lock); mutex_exit(&ztest_vdev_lock); } /* * Verify that we can't destroy an active pool, create an existing pool, * or create a pool with a bad vdev spec. */ void ztest_spa_create_destroy(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_opts_t *zo = &ztest_opts; spa_t *spa; nvlist_t *nvroot; if (zo->zo_mmp_test) return; /* * Attempt to create using a bad file. */ nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 0, 1); VERIFY3U(ENOENT, ==, spa_create("ztest_bad_file", nvroot, NULL, NULL, NULL)); fnvlist_free(nvroot); /* * Attempt to create using a bad mirror. */ nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 2, 1); VERIFY3U(ENOENT, ==, spa_create("ztest_bad_mirror", nvroot, NULL, NULL, NULL)); fnvlist_free(nvroot); /* * Attempt to create an existing pool. It shouldn't matter * what's in the nvroot; we should fail with EEXIST. */ (void) pthread_rwlock_rdlock(&ztest_name_lock); nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 0, 1); VERIFY3U(EEXIST, ==, spa_create(zo->zo_pool, nvroot, NULL, NULL, NULL)); fnvlist_free(nvroot); /* * We open a reference to the spa and then we try to export it * expecting one of the following errors: * * EBUSY * Because of the reference we just opened. * * ZFS_ERR_EXPORT_IN_PROGRESS * For the case that there is another ztest thread doing * an export concurrently. */ VERIFY0(spa_open(zo->zo_pool, &spa, FTAG)); int error = spa_destroy(zo->zo_pool); if (error != EBUSY && error != ZFS_ERR_EXPORT_IN_PROGRESS) { fatal(B_FALSE, "spa_destroy(%s) returned unexpected value %d", spa->spa_name, error); } spa_close(spa, FTAG); (void) pthread_rwlock_unlock(&ztest_name_lock); } /* * Start and then stop the MMP threads to ensure the startup and shutdown code * works properly. Actual protection and property-related code tested via ZTS. */ void ztest_mmp_enable_disable(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_opts_t *zo = &ztest_opts; spa_t *spa = ztest_spa; if (zo->zo_mmp_test) return; /* * Since enabling MMP involves setting a property, it could not be done * while the pool is suspended. */ if (spa_suspended(spa)) return; spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); mutex_enter(&spa->spa_props_lock); zfs_multihost_fail_intervals = 0; if (!spa_multihost(spa)) { spa->spa_multihost = B_TRUE; mmp_thread_start(spa); } mutex_exit(&spa->spa_props_lock); spa_config_exit(spa, SCL_CONFIG, FTAG); txg_wait_synced(spa_get_dsl(spa), 0); mmp_signal_all_threads(); txg_wait_synced(spa_get_dsl(spa), 0); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); mutex_enter(&spa->spa_props_lock); if (spa_multihost(spa)) { mmp_thread_stop(spa); spa->spa_multihost = B_FALSE; } mutex_exit(&spa->spa_props_lock); spa_config_exit(spa, SCL_CONFIG, FTAG); } void ztest_spa_upgrade(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa; uint64_t initial_version = SPA_VERSION_INITIAL; uint64_t version, newversion; nvlist_t *nvroot, *props; char *name; if (ztest_opts.zo_mmp_test) return; /* dRAID added after feature flags, skip upgrade test. */ if (strcmp(ztest_opts.zo_raid_type, VDEV_TYPE_DRAID) == 0) return; mutex_enter(&ztest_vdev_lock); name = kmem_asprintf("%s_upgrade", ztest_opts.zo_pool); /* * Clean up from previous runs. */ (void) spa_destroy(name); nvroot = make_vdev_root(NULL, NULL, name, ztest_opts.zo_vdev_size, 0, NULL, ztest_opts.zo_raid_children, ztest_opts.zo_mirrors, 1); /* * If we're configuring a RAIDZ device then make sure that the * initial version is capable of supporting that feature. */ switch (ztest_opts.zo_raid_parity) { case 0: case 1: initial_version = SPA_VERSION_INITIAL; break; case 2: initial_version = SPA_VERSION_RAIDZ2; break; case 3: initial_version = SPA_VERSION_RAIDZ3; break; } /* * Create a pool with a spa version that can be upgraded. Pick * a value between initial_version and SPA_VERSION_BEFORE_FEATURES. */ do { version = ztest_random_spa_version(initial_version); } while (version > SPA_VERSION_BEFORE_FEATURES); props = fnvlist_alloc(); fnvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), version); VERIFY0(spa_create(name, nvroot, props, NULL, NULL)); fnvlist_free(nvroot); fnvlist_free(props); VERIFY0(spa_open(name, &spa, FTAG)); VERIFY3U(spa_version(spa), ==, version); newversion = ztest_random_spa_version(version + 1); if (ztest_opts.zo_verbose >= 4) { (void) printf("upgrading spa version from " "%"PRIu64" to %"PRIu64"\n", version, newversion); } spa_upgrade(spa, newversion); VERIFY3U(spa_version(spa), >, version); VERIFY3U(spa_version(spa), ==, fnvlist_lookup_uint64(spa->spa_config, zpool_prop_to_name(ZPOOL_PROP_VERSION))); spa_close(spa, FTAG); kmem_strfree(name); mutex_exit(&ztest_vdev_lock); } static void ztest_spa_checkpoint(spa_t *spa) { ASSERT(MUTEX_HELD(&ztest_checkpoint_lock)); int error = spa_checkpoint(spa->spa_name); switch (error) { case 0: case ZFS_ERR_DEVRM_IN_PROGRESS: case ZFS_ERR_DISCARDING_CHECKPOINT: case ZFS_ERR_CHECKPOINT_EXISTS: break; case ENOSPC: ztest_record_enospc(FTAG); break; default: fatal(B_FALSE, "spa_checkpoint(%s) = %d", spa->spa_name, error); } } static void ztest_spa_discard_checkpoint(spa_t *spa) { ASSERT(MUTEX_HELD(&ztest_checkpoint_lock)); int error = spa_checkpoint_discard(spa->spa_name); switch (error) { case 0: case ZFS_ERR_DISCARDING_CHECKPOINT: case ZFS_ERR_NO_CHECKPOINT: break; default: fatal(B_FALSE, "spa_discard_checkpoint(%s) = %d", spa->spa_name, error); } } void ztest_spa_checkpoint_create_discard(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; mutex_enter(&ztest_checkpoint_lock); if (ztest_random(2) == 0) { ztest_spa_checkpoint(spa); } else { ztest_spa_discard_checkpoint(spa); } mutex_exit(&ztest_checkpoint_lock); } static vdev_t * vdev_lookup_by_path(vdev_t *vd, const char *path) { vdev_t *mvd; int c; if (vd->vdev_path != NULL && strcmp(path, vd->vdev_path) == 0) return (vd); for (c = 0; c < vd->vdev_children; c++) if ((mvd = vdev_lookup_by_path(vd->vdev_child[c], path)) != NULL) return (mvd); return (NULL); } static int spa_num_top_vdevs(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; ASSERT3U(spa_config_held(spa, SCL_VDEV, RW_READER), ==, SCL_VDEV); return (rvd->vdev_children); } /* * Verify that vdev_add() works as expected. */ void ztest_vdev_add_remove(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; uint64_t leaves; uint64_t guid; nvlist_t *nvroot; int error; if (ztest_opts.zo_mmp_test) return; mutex_enter(&ztest_vdev_lock); leaves = MAX(zs->zs_mirrors + zs->zs_splits, 1) * ztest_opts.zo_raid_children; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); ztest_shared->zs_vdev_next_leaf = spa_num_top_vdevs(spa) * leaves; /* * If we have slogs then remove them 1/4 of the time. */ if (spa_has_slogs(spa) && ztest_random(4) == 0) { metaslab_group_t *mg; /* * find the first real slog in log allocation class */ mg = spa_log_class(spa)->mc_allocator[0].mca_rotor; while (!mg->mg_vd->vdev_islog) mg = mg->mg_next; guid = mg->mg_vd->vdev_guid; spa_config_exit(spa, SCL_VDEV, FTAG); /* * We have to grab the zs_name_lock as writer to * prevent a race between removing a slog (dmu_objset_find) * and destroying a dataset. Removing the slog will * grab a reference on the dataset which may cause * dsl_destroy_head() to fail with EBUSY thus * leaving the dataset in an inconsistent state. */ pthread_rwlock_wrlock(&ztest_name_lock); error = spa_vdev_remove(spa, guid, B_FALSE); pthread_rwlock_unlock(&ztest_name_lock); switch (error) { case 0: case EEXIST: /* Generic zil_reset() error */ case EBUSY: /* Replay required */ case EACCES: /* Crypto key not loaded */ case ZFS_ERR_CHECKPOINT_EXISTS: case ZFS_ERR_DISCARDING_CHECKPOINT: break; default: fatal(B_FALSE, "spa_vdev_remove() = %d", error); } } else { spa_config_exit(spa, SCL_VDEV, FTAG); /* * Make 1/4 of the devices be log devices */ nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0, (ztest_random(4) == 0) ? "log" : NULL, ztest_opts.zo_raid_children, zs->zs_mirrors, 1); error = spa_vdev_add(spa, nvroot); fnvlist_free(nvroot); switch (error) { case 0: break; case ENOSPC: ztest_record_enospc("spa_vdev_add"); break; default: fatal(B_FALSE, "spa_vdev_add() = %d", error); } } mutex_exit(&ztest_vdev_lock); } void ztest_vdev_class_add(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; uint64_t leaves; nvlist_t *nvroot; const char *class = (ztest_random(2) == 0) ? VDEV_ALLOC_BIAS_SPECIAL : VDEV_ALLOC_BIAS_DEDUP; int error; /* * By default add a special vdev 50% of the time */ if ((ztest_opts.zo_special_vdevs == ZTEST_VDEV_CLASS_OFF) || (ztest_opts.zo_special_vdevs == ZTEST_VDEV_CLASS_RND && ztest_random(2) == 0)) { return; } mutex_enter(&ztest_vdev_lock); /* Only test with mirrors */ if (zs->zs_mirrors < 2) { mutex_exit(&ztest_vdev_lock); return; } /* requires feature@allocation_classes */ if (!spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)) { mutex_exit(&ztest_vdev_lock); return; } leaves = MAX(zs->zs_mirrors + zs->zs_splits, 1) * ztest_opts.zo_raid_children; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); ztest_shared->zs_vdev_next_leaf = spa_num_top_vdevs(spa) * leaves; spa_config_exit(spa, SCL_VDEV, FTAG); nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0, class, ztest_opts.zo_raid_children, zs->zs_mirrors, 1); error = spa_vdev_add(spa, nvroot); fnvlist_free(nvroot); if (error == ENOSPC) ztest_record_enospc("spa_vdev_add"); else if (error != 0) fatal(B_FALSE, "spa_vdev_add() = %d", error); /* * 50% of the time allow small blocks in the special class */ if (error == 0 && spa_special_class(spa)->mc_groups == 1 && ztest_random(2) == 0) { if (ztest_opts.zo_verbose >= 3) (void) printf("Enabling special VDEV small blocks\n"); (void) ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_SPECIAL_SMALL_BLOCKS, 32768, B_FALSE); } mutex_exit(&ztest_vdev_lock); if (ztest_opts.zo_verbose >= 3) { metaslab_class_t *mc; if (strcmp(class, VDEV_ALLOC_BIAS_SPECIAL) == 0) mc = spa_special_class(spa); else mc = spa_dedup_class(spa); (void) printf("Added a %s mirrored vdev (of %d)\n", class, (int)mc->mc_groups); } } /* * Verify that adding/removing aux devices (l2arc, hot spare) works as expected. */ void ztest_vdev_aux_add_remove(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; vdev_t *rvd = spa->spa_root_vdev; spa_aux_vdev_t *sav; const char *aux; char *path; uint64_t guid = 0; int error, ignore_err = 0; if (ztest_opts.zo_mmp_test) return; path = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); if (ztest_random(2) == 0) { sav = &spa->spa_spares; aux = ZPOOL_CONFIG_SPARES; } else { sav = &spa->spa_l2cache; aux = ZPOOL_CONFIG_L2CACHE; } mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); if (sav->sav_count != 0 && ztest_random(4) == 0) { /* * Pick a random device to remove. */ vdev_t *svd = sav->sav_vdevs[ztest_random(sav->sav_count)]; /* dRAID spares cannot be removed; try anyways to see ENOTSUP */ if (strstr(svd->vdev_path, VDEV_TYPE_DRAID) != NULL) ignore_err = ENOTSUP; guid = svd->vdev_guid; } else { /* * Find an unused device we can add. */ zs->zs_vdev_aux = 0; for (;;) { int c; (void) snprintf(path, MAXPATHLEN, ztest_aux_template, ztest_opts.zo_dir, ztest_opts.zo_pool, aux, zs->zs_vdev_aux); for (c = 0; c < sav->sav_count; c++) if (strcmp(sav->sav_vdevs[c]->vdev_path, path) == 0) break; if (c == sav->sav_count && vdev_lookup_by_path(rvd, path) == NULL) break; zs->zs_vdev_aux++; } } spa_config_exit(spa, SCL_VDEV, FTAG); if (guid == 0) { /* * Add a new device. */ nvlist_t *nvroot = make_vdev_root(NULL, aux, NULL, (ztest_opts.zo_vdev_size * 5) / 4, 0, NULL, 0, 0, 1); error = spa_vdev_add(spa, nvroot); switch (error) { case 0: break; default: fatal(B_FALSE, "spa_vdev_add(%p) = %d", nvroot, error); } fnvlist_free(nvroot); } else { /* * Remove an existing device. Sometimes, dirty its * vdev state first to make sure we handle removal * of devices that have pending state changes. */ if (ztest_random(2) == 0) (void) vdev_online(spa, guid, 0, NULL); error = spa_vdev_remove(spa, guid, B_FALSE); switch (error) { case 0: case EBUSY: case ZFS_ERR_CHECKPOINT_EXISTS: case ZFS_ERR_DISCARDING_CHECKPOINT: break; default: if (error != ignore_err) fatal(B_FALSE, "spa_vdev_remove(%"PRIu64") = %d", guid, error); } } mutex_exit(&ztest_vdev_lock); umem_free(path, MAXPATHLEN); } /* * split a pool if it has mirror tlvdevs */ void ztest_split_pool(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; vdev_t *rvd = spa->spa_root_vdev; nvlist_t *tree, **child, *config, *split, **schild; uint_t c, children, schildren = 0, lastlogid = 0; int error = 0; if (ztest_opts.zo_mmp_test) return; mutex_enter(&ztest_vdev_lock); /* ensure we have a usable config; mirrors of raidz aren't supported */ if (zs->zs_mirrors < 3 || ztest_opts.zo_raid_children > 1) { mutex_exit(&ztest_vdev_lock); return; } /* clean up the old pool, if any */ (void) spa_destroy("splitp"); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* generate a config from the existing config */ mutex_enter(&spa->spa_props_lock); tree = fnvlist_lookup_nvlist(spa->spa_config, ZPOOL_CONFIG_VDEV_TREE); mutex_exit(&spa->spa_props_lock); VERIFY0(nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &child, &children)); schild = malloc(rvd->vdev_children * sizeof (nvlist_t *)); for (c = 0; c < children; c++) { vdev_t *tvd = rvd->vdev_child[c]; nvlist_t **mchild; uint_t mchildren; if (tvd->vdev_islog || tvd->vdev_ops == &vdev_hole_ops) { schild[schildren] = fnvlist_alloc(); fnvlist_add_string(schild[schildren], ZPOOL_CONFIG_TYPE, VDEV_TYPE_HOLE); fnvlist_add_uint64(schild[schildren], ZPOOL_CONFIG_IS_HOLE, 1); if (lastlogid == 0) lastlogid = schildren; ++schildren; continue; } lastlogid = 0; VERIFY0(nvlist_lookup_nvlist_array(child[c], ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren)); schild[schildren++] = fnvlist_dup(mchild[0]); } /* OK, create a config that can be used to split */ split = fnvlist_alloc(); fnvlist_add_string(split, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT); fnvlist_add_nvlist_array(split, ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)schild, lastlogid != 0 ? lastlogid : schildren); config = fnvlist_alloc(); fnvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, split); for (c = 0; c < schildren; c++) fnvlist_free(schild[c]); free(schild); fnvlist_free(split); spa_config_exit(spa, SCL_VDEV, FTAG); (void) pthread_rwlock_wrlock(&ztest_name_lock); error = spa_vdev_split_mirror(spa, "splitp", config, NULL, B_FALSE); (void) pthread_rwlock_unlock(&ztest_name_lock); fnvlist_free(config); if (error == 0) { (void) printf("successful split - results:\n"); mutex_enter(&spa_namespace_lock); show_pool_stats(spa); show_pool_stats(spa_lookup("splitp")); mutex_exit(&spa_namespace_lock); ++zs->zs_splits; --zs->zs_mirrors; } mutex_exit(&ztest_vdev_lock); } /* * Verify that we can attach and detach devices. */ void ztest_vdev_attach_detach(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; spa_aux_vdev_t *sav = &spa->spa_spares; vdev_t *rvd = spa->spa_root_vdev; vdev_t *oldvd, *newvd, *pvd; nvlist_t *root; uint64_t leaves; uint64_t leaf, top; uint64_t ashift = ztest_get_ashift(); uint64_t oldguid, pguid; uint64_t oldsize, newsize; char *oldpath, *newpath; int replacing; int oldvd_has_siblings = B_FALSE; int newvd_is_spare = B_FALSE; int newvd_is_dspare = B_FALSE; int oldvd_is_log; int error, expected_error; if (ztest_opts.zo_mmp_test) return; oldpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); newpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); mutex_enter(&ztest_vdev_lock); leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raid_children; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * If a vdev is in the process of being removed, its removal may * finish while we are in progress, leading to an unexpected error * value. Don't bother trying to attach while we are in the middle * of removal. */ if (ztest_device_removal_active) { spa_config_exit(spa, SCL_ALL, FTAG); goto out; } /* * Decide whether to do an attach or a replace. */ replacing = ztest_random(2); /* * Pick a random top-level vdev. */ top = ztest_random_vdev_top(spa, B_TRUE); /* * Pick a random leaf within it. */ leaf = ztest_random(leaves); /* * Locate this vdev. */ oldvd = rvd->vdev_child[top]; /* pick a child from the mirror */ if (zs->zs_mirrors >= 1) { ASSERT3P(oldvd->vdev_ops, ==, &vdev_mirror_ops); ASSERT3U(oldvd->vdev_children, >=, zs->zs_mirrors); oldvd = oldvd->vdev_child[leaf / ztest_opts.zo_raid_children]; } /* pick a child out of the raidz group */ if (ztest_opts.zo_raid_children > 1) { if (strcmp(oldvd->vdev_ops->vdev_op_type, "raidz") == 0) ASSERT3P(oldvd->vdev_ops, ==, &vdev_raidz_ops); else ASSERT3P(oldvd->vdev_ops, ==, &vdev_draid_ops); ASSERT3U(oldvd->vdev_children, ==, ztest_opts.zo_raid_children); oldvd = oldvd->vdev_child[leaf % ztest_opts.zo_raid_children]; } /* * If we're already doing an attach or replace, oldvd may be a * mirror vdev -- in which case, pick a random child. */ while (oldvd->vdev_children != 0) { oldvd_has_siblings = B_TRUE; ASSERT3U(oldvd->vdev_children, >=, 2); oldvd = oldvd->vdev_child[ztest_random(oldvd->vdev_children)]; } oldguid = oldvd->vdev_guid; oldsize = vdev_get_min_asize(oldvd); oldvd_is_log = oldvd->vdev_top->vdev_islog; (void) strlcpy(oldpath, oldvd->vdev_path, MAXPATHLEN); pvd = oldvd->vdev_parent; pguid = pvd->vdev_guid; /* * If oldvd has siblings, then half of the time, detach it. Prior * to the detach the pool is scrubbed in order to prevent creating * unrepairable blocks as a result of the data corruption injection. */ if (oldvd_has_siblings && ztest_random(2) == 0) { spa_config_exit(spa, SCL_ALL, FTAG); error = ztest_scrub_impl(spa); if (error) goto out; error = spa_vdev_detach(spa, oldguid, pguid, B_FALSE); if (error != 0 && error != ENODEV && error != EBUSY && error != ENOTSUP && error != ZFS_ERR_CHECKPOINT_EXISTS && error != ZFS_ERR_DISCARDING_CHECKPOINT) fatal(B_FALSE, "detach (%s) returned %d", oldpath, error); goto out; } /* * For the new vdev, choose with equal probability between the two * standard paths (ending in either 'a' or 'b') or a random hot spare. */ if (sav->sav_count != 0 && ztest_random(3) == 0) { newvd = sav->sav_vdevs[ztest_random(sav->sav_count)]; newvd_is_spare = B_TRUE; if (newvd->vdev_ops == &vdev_draid_spare_ops) newvd_is_dspare = B_TRUE; (void) strlcpy(newpath, newvd->vdev_path, MAXPATHLEN); } else { (void) snprintf(newpath, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + leaf); if (ztest_random(2) == 0) newpath[strlen(newpath) - 1] = 'b'; newvd = vdev_lookup_by_path(rvd, newpath); } if (newvd) { /* * Reopen to ensure the vdev's asize field isn't stale. */ vdev_reopen(newvd); newsize = vdev_get_min_asize(newvd); } else { /* * Make newsize a little bigger or smaller than oldsize. * If it's smaller, the attach should fail. * If it's larger, and we're doing a replace, * we should get dynamic LUN growth when we're done. */ newsize = 10 * oldsize / (9 + ztest_random(3)); } /* * If pvd is not a mirror or root, the attach should fail with ENOTSUP, * unless it's a replace; in that case any non-replacing parent is OK. * * If newvd is already part of the pool, it should fail with EBUSY. * * If newvd is too small, it should fail with EOVERFLOW. * * If newvd is a distributed spare and it's being attached to a * dRAID which is not its parent it should fail with EINVAL. */ if (pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_root_ops && (!replacing || pvd->vdev_ops == &vdev_replacing_ops || pvd->vdev_ops == &vdev_spare_ops)) expected_error = ENOTSUP; else if (newvd_is_spare && (!replacing || oldvd_is_log)) expected_error = ENOTSUP; else if (newvd == oldvd) expected_error = replacing ? 0 : EBUSY; else if (vdev_lookup_by_path(rvd, newpath) != NULL) expected_error = EBUSY; else if (!newvd_is_dspare && newsize < oldsize) expected_error = EOVERFLOW; else if (ashift > oldvd->vdev_top->vdev_ashift) expected_error = EDOM; else if (newvd_is_dspare && pvd != vdev_draid_spare_get_parent(newvd)) expected_error = ENOTSUP; else expected_error = 0; spa_config_exit(spa, SCL_ALL, FTAG); /* * Build the nvlist describing newpath. */ root = make_vdev_root(newpath, NULL, NULL, newvd == NULL ? newsize : 0, ashift, NULL, 0, 0, 1); /* * When supported select either a healing or sequential resilver. */ boolean_t rebuilding = B_FALSE; if (pvd->vdev_ops == &vdev_mirror_ops || pvd->vdev_ops == &vdev_root_ops) { rebuilding = !!ztest_random(2); } error = spa_vdev_attach(spa, oldguid, root, replacing, rebuilding); fnvlist_free(root); /* * If our parent was the replacing vdev, but the replace completed, * then instead of failing with ENOTSUP we may either succeed, * fail with ENODEV, or fail with EOVERFLOW. */ if (expected_error == ENOTSUP && (error == 0 || error == ENODEV || error == EOVERFLOW)) expected_error = error; /* * If someone grew the LUN, the replacement may be too small. */ if (error == EOVERFLOW || error == EBUSY) expected_error = error; if (error == ZFS_ERR_CHECKPOINT_EXISTS || error == ZFS_ERR_DISCARDING_CHECKPOINT || error == ZFS_ERR_RESILVER_IN_PROGRESS || error == ZFS_ERR_REBUILD_IN_PROGRESS) expected_error = error; if (error != expected_error && expected_error != EBUSY) { fatal(B_FALSE, "attach (%s %"PRIu64", %s %"PRIu64", %d) " "returned %d, expected %d", oldpath, oldsize, newpath, newsize, replacing, error, expected_error); } out: mutex_exit(&ztest_vdev_lock); umem_free(oldpath, MAXPATHLEN); umem_free(newpath, MAXPATHLEN); } void ztest_device_removal(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; vdev_t *vd; uint64_t guid; int error; mutex_enter(&ztest_vdev_lock); if (ztest_device_removal_active) { mutex_exit(&ztest_vdev_lock); return; } /* * Remove a random top-level vdev and wait for removal to finish. */ spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); vd = vdev_lookup_top(spa, ztest_random_vdev_top(spa, B_FALSE)); guid = vd->vdev_guid; spa_config_exit(spa, SCL_VDEV, FTAG); error = spa_vdev_remove(spa, guid, B_FALSE); if (error == 0) { ztest_device_removal_active = B_TRUE; mutex_exit(&ztest_vdev_lock); /* * spa->spa_vdev_removal is created in a sync task that * is initiated via dsl_sync_task_nowait(). Since the * task may not run before spa_vdev_remove() returns, we * must wait at least 1 txg to ensure that the removal * struct has been created. */ txg_wait_synced(spa_get_dsl(spa), 0); while (spa->spa_removing_phys.sr_state == DSS_SCANNING) txg_wait_synced(spa_get_dsl(spa), 0); } else { mutex_exit(&ztest_vdev_lock); return; } /* * The pool needs to be scrubbed after completing device removal. * Failure to do so may result in checksum errors due to the * strategy employed by ztest_fault_inject() when selecting which * offset are redundant and can be damaged. */ error = spa_scan(spa, POOL_SCAN_SCRUB); if (error == 0) { while (dsl_scan_scrubbing(spa_get_dsl(spa))) txg_wait_synced(spa_get_dsl(spa), 0); } mutex_enter(&ztest_vdev_lock); ztest_device_removal_active = B_FALSE; mutex_exit(&ztest_vdev_lock); } /* * Callback function which expands the physical size of the vdev. */ static vdev_t * grow_vdev(vdev_t *vd, void *arg) { spa_t *spa __maybe_unused = vd->vdev_spa; size_t *newsize = arg; size_t fsize; int fd; ASSERT3S(spa_config_held(spa, SCL_STATE, RW_READER), ==, SCL_STATE); ASSERT(vd->vdev_ops->vdev_op_leaf); if ((fd = open(vd->vdev_path, O_RDWR)) == -1) return (vd); fsize = lseek(fd, 0, SEEK_END); VERIFY0(ftruncate(fd, *newsize)); if (ztest_opts.zo_verbose >= 6) { (void) printf("%s grew from %lu to %lu bytes\n", vd->vdev_path, (ulong_t)fsize, (ulong_t)*newsize); } (void) close(fd); return (NULL); } /* * Callback function which expands a given vdev by calling vdev_online(). */ static vdev_t * online_vdev(vdev_t *vd, void *arg) { (void) arg; spa_t *spa = vd->vdev_spa; vdev_t *tvd = vd->vdev_top; uint64_t guid = vd->vdev_guid; uint64_t generation = spa->spa_config_generation + 1; vdev_state_t newstate = VDEV_STATE_UNKNOWN; int error; ASSERT3S(spa_config_held(spa, SCL_STATE, RW_READER), ==, SCL_STATE); ASSERT(vd->vdev_ops->vdev_op_leaf); /* Calling vdev_online will initialize the new metaslabs */ spa_config_exit(spa, SCL_STATE, spa); error = vdev_online(spa, guid, ZFS_ONLINE_EXPAND, &newstate); spa_config_enter(spa, SCL_STATE, spa, RW_READER); /* * If vdev_online returned an error or the underlying vdev_open * failed then we abort the expand. The only way to know that * vdev_open fails is by checking the returned newstate. */ if (error || newstate != VDEV_STATE_HEALTHY) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Unable to expand vdev, state %u, " "error %d\n", newstate, error); } return (vd); } ASSERT3U(newstate, ==, VDEV_STATE_HEALTHY); /* * Since we dropped the lock we need to ensure that we're * still talking to the original vdev. It's possible this * vdev may have been detached/replaced while we were * trying to online it. */ if (generation != spa->spa_config_generation) { if (ztest_opts.zo_verbose >= 5) { (void) printf("vdev configuration has changed, " "guid %"PRIu64", state %"PRIu64", " "expected gen %"PRIu64", got gen %"PRIu64"\n", guid, tvd->vdev_state, generation, spa->spa_config_generation); } return (vd); } return (NULL); } /* * Traverse the vdev tree calling the supplied function. * We continue to walk the tree until we either have walked all * children or we receive a non-NULL return from the callback. * If a NULL callback is passed, then we just return back the first * leaf vdev we encounter. */ static vdev_t * vdev_walk_tree(vdev_t *vd, vdev_t *(*func)(vdev_t *, void *), void *arg) { uint_t c; if (vd->vdev_ops->vdev_op_leaf) { if (func == NULL) return (vd); else return (func(vd, arg)); } for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; if ((cvd = vdev_walk_tree(cvd, func, arg)) != NULL) return (cvd); } return (NULL); } /* * Verify that dynamic LUN growth works as expected. */ void ztest_vdev_LUN_growth(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; vdev_t *vd, *tvd; metaslab_class_t *mc; metaslab_group_t *mg; size_t psize, newsize; uint64_t top; uint64_t old_class_space, new_class_space, old_ms_count, new_ms_count; mutex_enter(&ztest_checkpoint_lock); mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_STATE, spa, RW_READER); /* * If there is a vdev removal in progress, it could complete while * we are running, in which case we would not be able to verify * that the metaslab_class space increased (because it decreases * when the device removal completes). */ if (ztest_device_removal_active) { spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); mutex_exit(&ztest_checkpoint_lock); return; } top = ztest_random_vdev_top(spa, B_TRUE); tvd = spa->spa_root_vdev->vdev_child[top]; mg = tvd->vdev_mg; mc = mg->mg_class; old_ms_count = tvd->vdev_ms_count; old_class_space = metaslab_class_get_space(mc); /* * Determine the size of the first leaf vdev associated with * our top-level device. */ vd = vdev_walk_tree(tvd, NULL, NULL); ASSERT3P(vd, !=, NULL); ASSERT(vd->vdev_ops->vdev_op_leaf); psize = vd->vdev_psize; /* * We only try to expand the vdev if it's healthy, less than 4x its * original size, and it has a valid psize. */ if (tvd->vdev_state != VDEV_STATE_HEALTHY || psize == 0 || psize >= 4 * ztest_opts.zo_vdev_size) { spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); mutex_exit(&ztest_checkpoint_lock); return; } ASSERT3U(psize, >, 0); newsize = psize + MAX(psize / 8, SPA_MAXBLOCKSIZE); ASSERT3U(newsize, >, psize); if (ztest_opts.zo_verbose >= 6) { (void) printf("Expanding LUN %s from %lu to %lu\n", vd->vdev_path, (ulong_t)psize, (ulong_t)newsize); } /* * Growing the vdev is a two step process: * 1). expand the physical size (i.e. relabel) * 2). online the vdev to create the new metaslabs */ if (vdev_walk_tree(tvd, grow_vdev, &newsize) != NULL || vdev_walk_tree(tvd, online_vdev, NULL) != NULL || tvd->vdev_state != VDEV_STATE_HEALTHY) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Could not expand LUN because " "the vdev configuration changed.\n"); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); mutex_exit(&ztest_checkpoint_lock); return; } spa_config_exit(spa, SCL_STATE, spa); /* * Expanding the LUN will update the config asynchronously, * thus we must wait for the async thread to complete any * pending tasks before proceeding. */ for (;;) { boolean_t done; mutex_enter(&spa->spa_async_lock); done = (spa->spa_async_thread == NULL && !spa->spa_async_tasks); mutex_exit(&spa->spa_async_lock); if (done) break; txg_wait_synced(spa_get_dsl(spa), 0); (void) poll(NULL, 0, 100); } spa_config_enter(spa, SCL_STATE, spa, RW_READER); tvd = spa->spa_root_vdev->vdev_child[top]; new_ms_count = tvd->vdev_ms_count; new_class_space = metaslab_class_get_space(mc); if (tvd->vdev_mg != mg || mg->mg_class != mc) { if (ztest_opts.zo_verbose >= 5) { (void) printf("Could not verify LUN expansion due to " "intervening vdev offline or remove.\n"); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); mutex_exit(&ztest_checkpoint_lock); return; } /* * Make sure we were able to grow the vdev. */ if (new_ms_count <= old_ms_count) { fatal(B_FALSE, "LUN expansion failed: ms_count %"PRIu64" < %"PRIu64"\n", old_ms_count, new_ms_count); } /* * Make sure we were able to grow the pool. */ if (new_class_space <= old_class_space) { fatal(B_FALSE, "LUN expansion failed: class_space %"PRIu64" < %"PRIu64"\n", old_class_space, new_class_space); } if (ztest_opts.zo_verbose >= 5) { char oldnumbuf[NN_NUMBUF_SZ], newnumbuf[NN_NUMBUF_SZ]; nicenum(old_class_space, oldnumbuf, sizeof (oldnumbuf)); nicenum(new_class_space, newnumbuf, sizeof (newnumbuf)); (void) printf("%s grew from %s to %s\n", spa->spa_name, oldnumbuf, newnumbuf); } spa_config_exit(spa, SCL_STATE, spa); mutex_exit(&ztest_vdev_lock); mutex_exit(&ztest_checkpoint_lock); } /* * Verify that dmu_objset_{create,destroy,open,close} work as expected. */ static void ztest_objset_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { (void) arg, (void) cr; /* * Create the objects common to all ztest datasets. */ VERIFY0(zap_create_claim(os, ZTEST_DIROBJ, DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx)); } static int ztest_dataset_create(char *dsname) { int err; uint64_t rand; dsl_crypto_params_t *dcp = NULL; /* * 50% of the time, we create encrypted datasets * using a random cipher suite and a hard-coded * wrapping key. */ rand = ztest_random(2); if (rand != 0) { nvlist_t *crypto_args = fnvlist_alloc(); nvlist_t *props = fnvlist_alloc(); /* slight bias towards the default cipher suite */ rand = ztest_random(ZIO_CRYPT_FUNCTIONS); if (rand < ZIO_CRYPT_AES_128_CCM) rand = ZIO_CRYPT_ON; fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_ENCRYPTION), rand); fnvlist_add_uint8_array(crypto_args, "wkeydata", (uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN); /* * These parameters aren't really used by the kernel. They * are simply stored so that userspace knows how to load * the wrapping key. */ fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_KEYFORMAT), ZFS_KEYFORMAT_RAW); fnvlist_add_string(props, zfs_prop_to_name(ZFS_PROP_KEYLOCATION), "prompt"); fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_PBKDF2_SALT), 0ULL); fnvlist_add_uint64(props, zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), 0ULL); VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, props, crypto_args, &dcp)); /* * Cycle through all available encryption implementations * to verify interoperability. */ VERIFY0(gcm_impl_set("cycle")); VERIFY0(aes_impl_set("cycle")); fnvlist_free(crypto_args); fnvlist_free(props); } err = dmu_objset_create(dsname, DMU_OST_OTHER, 0, dcp, ztest_objset_create_cb, NULL); dsl_crypto_params_free(dcp, !!err); rand = ztest_random(100); if (err || rand < 80) return (err); if (ztest_opts.zo_verbose >= 5) (void) printf("Setting dataset %s to sync always\n", dsname); return (ztest_dsl_prop_set_uint64(dsname, ZFS_PROP_SYNC, ZFS_SYNC_ALWAYS, B_FALSE)); } static int ztest_objset_destroy_cb(const char *name, void *arg) { (void) arg; objset_t *os; dmu_object_info_t doi; int error; /* * Verify that the dataset contains a directory object. */ VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE, B_TRUE, FTAG, &os)); error = dmu_object_info(os, ZTEST_DIROBJ, &doi); if (error != ENOENT) { /* We could have crashed in the middle of destroying it */ ASSERT0(error); ASSERT3U(doi.doi_type, ==, DMU_OT_ZAP_OTHER); ASSERT3S(doi.doi_physical_blocks_512, >=, 0); } dmu_objset_disown(os, B_TRUE, FTAG); /* * Destroy the dataset. */ if (strchr(name, '@') != NULL) { error = dsl_destroy_snapshot(name, B_TRUE); if (error != ECHRNG) { /* * The program was executed, but encountered a runtime * error, such as insufficient slop, or a hold on the * dataset. */ ASSERT0(error); } } else { error = dsl_destroy_head(name); if (error == ENOSPC) { /* There could be checkpoint or insufficient slop */ ztest_record_enospc(FTAG); } else if (error != EBUSY) { /* There could be a hold on this dataset */ ASSERT0(error); } } return (0); } static boolean_t ztest_snapshot_create(char *osname, uint64_t id) { char snapname[ZFS_MAX_DATASET_NAME_LEN]; int error; (void) snprintf(snapname, sizeof (snapname), "%"PRIu64"", id); error = dmu_objset_snapshot_one(osname, snapname); if (error == ENOSPC) { ztest_record_enospc(FTAG); return (B_FALSE); } if (error != 0 && error != EEXIST) { fatal(B_FALSE, "ztest_snapshot_create(%s@%s) = %d", osname, snapname, error); } return (B_TRUE); } static boolean_t ztest_snapshot_destroy(char *osname, uint64_t id) { char snapname[ZFS_MAX_DATASET_NAME_LEN]; int error; (void) snprintf(snapname, sizeof (snapname), "%s@%"PRIu64"", osname, id); error = dsl_destroy_snapshot(snapname, B_FALSE); if (error != 0 && error != ENOENT) fatal(B_FALSE, "ztest_snapshot_destroy(%s) = %d", snapname, error); return (B_TRUE); } void ztest_dmu_objset_create_destroy(ztest_ds_t *zd, uint64_t id) { (void) zd; ztest_ds_t *zdtmp; int iters; int error; objset_t *os, *os2; char name[ZFS_MAX_DATASET_NAME_LEN]; zilog_t *zilog; int i; zdtmp = umem_alloc(sizeof (ztest_ds_t), UMEM_NOFAIL); (void) pthread_rwlock_rdlock(&ztest_name_lock); (void) snprintf(name, sizeof (name), "%s/temp_%"PRIu64"", ztest_opts.zo_pool, id); /* * If this dataset exists from a previous run, process its replay log * half of the time. If we don't replay it, then dsl_destroy_head() * (invoked from ztest_objset_destroy_cb()) should just throw it away. */ if (ztest_random(2) == 0 && ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os) == 0) { ztest_zd_init(zdtmp, NULL, os); zil_replay(os, zdtmp, ztest_replay_vector); ztest_zd_fini(zdtmp); dmu_objset_disown(os, B_TRUE, FTAG); } /* * There may be an old instance of the dataset we're about to * create lying around from a previous run. If so, destroy it * and all of its snapshots. */ (void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL, DS_FIND_CHILDREN | DS_FIND_SNAPSHOTS); /* * Verify that the destroyed dataset is no longer in the namespace. */ VERIFY3U(ENOENT, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE, B_TRUE, FTAG, &os)); /* * Verify that we can create a new dataset. */ error = ztest_dataset_create(name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_objset_create(%s) = %d", name, error); } VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os)); ztest_zd_init(zdtmp, NULL, os); /* * Open the intent log for it. */ zilog = zil_open(os, ztest_get_data, NULL); /* * Put some objects in there, do a little I/O to them, * and randomly take a couple of snapshots along the way. */ iters = ztest_random(5); for (i = 0; i < iters; i++) { ztest_dmu_object_alloc_free(zdtmp, id); if (ztest_random(iters) == 0) (void) ztest_snapshot_create(name, i); } /* * Verify that we cannot create an existing dataset. */ VERIFY3U(EEXIST, ==, dmu_objset_create(name, DMU_OST_OTHER, 0, NULL, NULL, NULL)); /* * Verify that we can hold an objset that is also owned. */ VERIFY0(dmu_objset_hold(name, FTAG, &os2)); dmu_objset_rele(os2, FTAG); /* * Verify that we cannot own an objset that is already owned. */ VERIFY3U(EBUSY, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, FTAG, &os2)); zil_close(zilog); dmu_objset_disown(os, B_TRUE, FTAG); ztest_zd_fini(zdtmp); out: (void) pthread_rwlock_unlock(&ztest_name_lock); umem_free(zdtmp, sizeof (ztest_ds_t)); } /* * Verify that dmu_snapshot_{create,destroy,open,close} work as expected. */ void ztest_dmu_snapshot_create_destroy(ztest_ds_t *zd, uint64_t id) { (void) pthread_rwlock_rdlock(&ztest_name_lock); (void) ztest_snapshot_destroy(zd->zd_name, id); (void) ztest_snapshot_create(zd->zd_name, id); (void) pthread_rwlock_unlock(&ztest_name_lock); } /* * Cleanup non-standard snapshots and clones. */ static void ztest_dsl_dataset_cleanup(char *osname, uint64_t id) { char *snap1name; char *clone1name; char *snap2name; char *clone2name; char *snap3name; int error; snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); (void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%"PRIu64"", osname, id); (void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%"PRIu64"", osname, id); (void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%"PRIu64"", clone1name, id); (void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%"PRIu64"", osname, id); (void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%"PRIu64"", clone1name, id); error = dsl_destroy_head(clone2name); if (error && error != ENOENT) fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clone2name, error); error = dsl_destroy_snapshot(snap3name, B_FALSE); if (error && error != ENOENT) fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d", snap3name, error); error = dsl_destroy_snapshot(snap2name, B_FALSE); if (error && error != ENOENT) fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d", snap2name, error); error = dsl_destroy_head(clone1name); if (error && error != ENOENT) fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clone1name, error); error = dsl_destroy_snapshot(snap1name, B_FALSE); if (error && error != ENOENT) fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d", snap1name, error); umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN); } /* * Verify dsl_dataset_promote handles EBUSY */ void ztest_dsl_dataset_promote_busy(ztest_ds_t *zd, uint64_t id) { objset_t *os; char *snap1name; char *clone1name; char *snap2name; char *clone2name; char *snap3name; char *osname = zd->zd_name; int error; snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL); (void) pthread_rwlock_rdlock(&ztest_name_lock); ztest_dsl_dataset_cleanup(osname, id); (void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%"PRIu64"", osname, id); (void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%"PRIu64"", osname, id); (void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%"PRIu64"", clone1name, id); (void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%"PRIu64"", osname, id); (void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%"PRIu64"", clone1name, id); error = dmu_objset_snapshot_one(osname, strchr(snap1name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_take_snapshot(%s) = %d", snap1name, error); } error = dmu_objset_clone(clone1name, snap1name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_objset_create(%s) = %d", clone1name, error); } error = dmu_objset_snapshot_one(clone1name, strchr(snap2name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_open_snapshot(%s) = %d", snap2name, error); } error = dmu_objset_snapshot_one(clone1name, strchr(snap3name, '@') + 1); if (error && error != EEXIST) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_open_snapshot(%s) = %d", snap3name, error); } error = dmu_objset_clone(clone2name, snap3name); if (error) { if (error == ENOSPC) { ztest_record_enospc(FTAG); goto out; } fatal(B_FALSE, "dmu_objset_create(%s) = %d", clone2name, error); } error = ztest_dmu_objset_own(snap2name, DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os); if (error) fatal(B_FALSE, "dmu_objset_own(%s) = %d", snap2name, error); error = dsl_dataset_promote(clone2name, NULL); if (error == ENOSPC) { dmu_objset_disown(os, B_TRUE, FTAG); ztest_record_enospc(FTAG); goto out; } if (error != EBUSY) fatal(B_FALSE, "dsl_dataset_promote(%s), %d, not EBUSY", clone2name, error); dmu_objset_disown(os, B_TRUE, FTAG); out: ztest_dsl_dataset_cleanup(osname, id); (void) pthread_rwlock_unlock(&ztest_name_lock); umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN); umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN); } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 4 /* * Verify that dmu_object_{alloc,free} work as expected. */ void ztest_dmu_object_alloc_free(ztest_ds_t *zd, uint64_t id) { ztest_od_t *od; int batchsize; int size; int b; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); batchsize = OD_ARRAY_SIZE; for (b = 0; b < batchsize; b++) ztest_od_init(od + b, id, FTAG, b, DMU_OT_UINT64_OTHER, 0, 0, 0); /* * Destroy the previous batch of objects, create a new batch, * and do some I/O on the new objects. */ if (ztest_object_init(zd, od, size, B_TRUE) != 0) return; while (ztest_random(4 * batchsize) != 0) ztest_io(zd, od[ztest_random(batchsize)].od_object, ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); umem_free(od, size); } /* * Rewind the global allocator to verify object allocation backfilling. */ void ztest_dmu_object_next_chunk(ztest_ds_t *zd, uint64_t id) { (void) id; objset_t *os = zd->zd_os; uint_t dnodes_per_chunk = 1 << dmu_object_alloc_chunk_shift; uint64_t object; /* * Rewind the global allocator randomly back to a lower object number * to force backfilling and reclamation of recently freed dnodes. */ mutex_enter(&os->os_obj_lock); object = ztest_random(os->os_obj_next_chunk); os->os_obj_next_chunk = P2ALIGN(object, dnodes_per_chunk); mutex_exit(&os->os_obj_lock); } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 2 /* * Verify that dmu_{read,write} work as expected. */ void ztest_dmu_read_write(ztest_ds_t *zd, uint64_t id) { int size; ztest_od_t *od; objset_t *os = zd->zd_os; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); dmu_tx_t *tx; int freeit, error; uint64_t i, n, s, txg; bufwad_t *packbuf, *bigbuf, *pack, *bigH, *bigT; uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize; uint64_t chunksize = (1000 + ztest_random(1000)) * sizeof (uint64_t); uint64_t regions = 997; uint64_t stride = 123456789ULL; uint64_t width = 40; int free_percent = 5; /* * This test uses two objects, packobj and bigobj, that are always * updated together (i.e. in the same tx) so that their contents are * in sync and can be compared. Their contents relate to each other * in a simple way: packobj is a dense array of 'bufwad' structures, * while bigobj is a sparse array of the same bufwads. Specifically, * for any index n, there are three bufwads that should be identical: * * packobj, at offset n * sizeof (bufwad_t) * bigobj, at the head of the nth chunk * bigobj, at the tail of the nth chunk * * The chunk size is arbitrary. It doesn't have to be a power of two, * and it doesn't have any relation to the object blocksize. * The only requirement is that it can hold at least two bufwads. * * Normally, we write the bufwad to each of these locations. * However, free_percent of the time we instead write zeroes to * packobj and perform a dmu_free_range() on bigobj. By comparing * bigobj to packobj, we can verify that the DMU is correctly * tracking which parts of an object are allocated and free, * and that the contents of the allocated blocks are correct. */ /* * Read the directory info. If it's the first time, set things up. */ ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, chunksize); ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0, chunksize); if (ztest_object_init(zd, od, size, B_FALSE) != 0) { umem_free(od, size); return; } bigobj = od[0].od_object; packobj = od[1].od_object; chunksize = od[0].od_gen; ASSERT3U(chunksize, ==, od[1].od_gen); /* * Prefetch a random chunk of the big object. * Our aim here is to get some async reads in flight * for blocks that we may free below; the DMU should * handle this race correctly. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(2 * width - 1); dmu_prefetch(os, bigobj, 0, n * chunksize, s * chunksize, ZIO_PRIORITY_SYNC_READ); /* * Pick a random index and compute the offsets into packobj and bigobj. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(width - 1); packoff = n * sizeof (bufwad_t); packsize = s * sizeof (bufwad_t); bigoff = n * chunksize; bigsize = s * chunksize; packbuf = umem_alloc(packsize, UMEM_NOFAIL); bigbuf = umem_alloc(bigsize, UMEM_NOFAIL); /* * free_percent of the time, free a range of bigobj rather than * overwriting it. */ freeit = (ztest_random(100) < free_percent); /* * Read the current contents of our objects. */ error = dmu_read(os, packobj, packoff, packsize, packbuf, DMU_READ_PREFETCH); ASSERT0(error); error = dmu_read(os, bigobj, bigoff, bigsize, bigbuf, DMU_READ_PREFETCH); ASSERT0(error); /* * Get a tx for the mods to both packobj and bigobj. */ tx = dmu_tx_create(os); dmu_tx_hold_write(tx, packobj, packoff, packsize); if (freeit) dmu_tx_hold_free(tx, bigobj, bigoff, bigsize); else dmu_tx_hold_write(tx, bigobj, bigoff, bigsize); /* This accounts for setting the checksum/compression. */ dmu_tx_hold_bonus(tx, bigobj); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(od, size); return; } enum zio_checksum cksum; do { cksum = (enum zio_checksum) ztest_random_dsl_prop(ZFS_PROP_CHECKSUM); } while (cksum >= ZIO_CHECKSUM_LEGACY_FUNCTIONS); dmu_object_set_checksum(os, bigobj, cksum, tx); enum zio_compress comp; do { comp = (enum zio_compress) ztest_random_dsl_prop(ZFS_PROP_COMPRESSION); } while (comp >= ZIO_COMPRESS_LEGACY_FUNCTIONS); dmu_object_set_compress(os, bigobj, comp, tx); /* * For each index from n to n + s, verify that the existing bufwad * in packobj matches the bufwads at the head and tail of the * corresponding chunk in bigobj. Then update all three bufwads * with the new values we want to write out. */ for (i = 0; i < s; i++) { /* LINTED */ pack = (bufwad_t *)((char *)packbuf + i * sizeof (bufwad_t)); /* LINTED */ bigH = (bufwad_t *)((char *)bigbuf + i * chunksize); /* LINTED */ bigT = (bufwad_t *)((char *)bigH + chunksize) - 1; ASSERT3U((uintptr_t)bigH - (uintptr_t)bigbuf, <, bigsize); ASSERT3U((uintptr_t)bigT - (uintptr_t)bigbuf, <, bigsize); if (pack->bw_txg > txg) fatal(B_FALSE, "future leak: got %"PRIx64", open txg is %"PRIx64"", pack->bw_txg, txg); if (pack->bw_data != 0 && pack->bw_index != n + i) fatal(B_FALSE, "wrong index: " "got %"PRIx64", wanted %"PRIx64"+%"PRIx64"", pack->bw_index, n, i); if (memcmp(pack, bigH, sizeof (bufwad_t)) != 0) fatal(B_FALSE, "pack/bigH mismatch in %p/%p", pack, bigH); if (memcmp(pack, bigT, sizeof (bufwad_t)) != 0) fatal(B_FALSE, "pack/bigT mismatch in %p/%p", pack, bigT); if (freeit) { memset(pack, 0, sizeof (bufwad_t)); } else { pack->bw_index = n + i; pack->bw_txg = txg; pack->bw_data = 1 + ztest_random(-2ULL); } *bigH = *pack; *bigT = *pack; } /* * We've verified all the old bufwads, and made new ones. * Now write them out. */ dmu_write(os, packobj, packoff, packsize, packbuf, tx); if (freeit) { if (ztest_opts.zo_verbose >= 7) { (void) printf("freeing offset %"PRIx64" size %"PRIx64"" " txg %"PRIx64"\n", bigoff, bigsize, txg); } VERIFY0(dmu_free_range(os, bigobj, bigoff, bigsize, tx)); } else { if (ztest_opts.zo_verbose >= 7) { (void) printf("writing offset %"PRIx64" size %"PRIx64"" " txg %"PRIx64"\n", bigoff, bigsize, txg); } dmu_write(os, bigobj, bigoff, bigsize, bigbuf, tx); } dmu_tx_commit(tx); /* * Sanity check the stuff we just wrote. */ { void *packcheck = umem_alloc(packsize, UMEM_NOFAIL); void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL); VERIFY0(dmu_read(os, packobj, packoff, packsize, packcheck, DMU_READ_PREFETCH)); VERIFY0(dmu_read(os, bigobj, bigoff, bigsize, bigcheck, DMU_READ_PREFETCH)); ASSERT0(memcmp(packbuf, packcheck, packsize)); ASSERT0(memcmp(bigbuf, bigcheck, bigsize)); umem_free(packcheck, packsize); umem_free(bigcheck, bigsize); } umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(od, size); } static void compare_and_update_pbbufs(uint64_t s, bufwad_t *packbuf, bufwad_t *bigbuf, uint64_t bigsize, uint64_t n, uint64_t chunksize, uint64_t txg) { uint64_t i; bufwad_t *pack; bufwad_t *bigH; bufwad_t *bigT; /* * For each index from n to n + s, verify that the existing bufwad * in packobj matches the bufwads at the head and tail of the * corresponding chunk in bigobj. Then update all three bufwads * with the new values we want to write out. */ for (i = 0; i < s; i++) { /* LINTED */ pack = (bufwad_t *)((char *)packbuf + i * sizeof (bufwad_t)); /* LINTED */ bigH = (bufwad_t *)((char *)bigbuf + i * chunksize); /* LINTED */ bigT = (bufwad_t *)((char *)bigH + chunksize) - 1; ASSERT3U((uintptr_t)bigH - (uintptr_t)bigbuf, <, bigsize); ASSERT3U((uintptr_t)bigT - (uintptr_t)bigbuf, <, bigsize); if (pack->bw_txg > txg) fatal(B_FALSE, "future leak: got %"PRIx64", open txg is %"PRIx64"", pack->bw_txg, txg); if (pack->bw_data != 0 && pack->bw_index != n + i) fatal(B_FALSE, "wrong index: " "got %"PRIx64", wanted %"PRIx64"+%"PRIx64"", pack->bw_index, n, i); if (memcmp(pack, bigH, sizeof (bufwad_t)) != 0) fatal(B_FALSE, "pack/bigH mismatch in %p/%p", pack, bigH); if (memcmp(pack, bigT, sizeof (bufwad_t)) != 0) fatal(B_FALSE, "pack/bigT mismatch in %p/%p", pack, bigT); pack->bw_index = n + i; pack->bw_txg = txg; pack->bw_data = 1 + ztest_random(-2ULL); *bigH = *pack; *bigT = *pack; } } #undef OD_ARRAY_SIZE #define OD_ARRAY_SIZE 2 void ztest_dmu_read_write_zcopy(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; dmu_tx_t *tx; uint64_t i; int error; int size; uint64_t n, s, txg; bufwad_t *packbuf, *bigbuf; uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize; uint64_t blocksize = ztest_random_blocksize(); uint64_t chunksize = blocksize; uint64_t regions = 997; uint64_t stride = 123456789ULL; uint64_t width = 9; dmu_buf_t *bonus_db; arc_buf_t **bigbuf_arcbufs; dmu_object_info_t doi; size = sizeof (ztest_od_t) * OD_ARRAY_SIZE; od = umem_alloc(size, UMEM_NOFAIL); /* * This test uses two objects, packobj and bigobj, that are always * updated together (i.e. in the same tx) so that their contents are * in sync and can be compared. Their contents relate to each other * in a simple way: packobj is a dense array of 'bufwad' structures, * while bigobj is a sparse array of the same bufwads. Specifically, * for any index n, there are three bufwads that should be identical: * * packobj, at offset n * sizeof (bufwad_t) * bigobj, at the head of the nth chunk * bigobj, at the tail of the nth chunk * * The chunk size is set equal to bigobj block size so that * dmu_assign_arcbuf_by_dbuf() can be tested for object updates. */ /* * Read the directory info. If it's the first time, set things up. */ ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0); ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0, chunksize); if (ztest_object_init(zd, od, size, B_FALSE) != 0) { umem_free(od, size); return; } bigobj = od[0].od_object; packobj = od[1].od_object; blocksize = od[0].od_blocksize; chunksize = blocksize; ASSERT3U(chunksize, ==, od[1].od_gen); VERIFY0(dmu_object_info(os, bigobj, &doi)); VERIFY(ISP2(doi.doi_data_block_size)); VERIFY3U(chunksize, ==, doi.doi_data_block_size); VERIFY3U(chunksize, >=, 2 * sizeof (bufwad_t)); /* * Pick a random index and compute the offsets into packobj and bigobj. */ n = ztest_random(regions) * stride + ztest_random(width); s = 1 + ztest_random(width - 1); packoff = n * sizeof (bufwad_t); packsize = s * sizeof (bufwad_t); bigoff = n * chunksize; bigsize = s * chunksize; packbuf = umem_zalloc(packsize, UMEM_NOFAIL); bigbuf = umem_zalloc(bigsize, UMEM_NOFAIL); VERIFY0(dmu_bonus_hold(os, bigobj, FTAG, &bonus_db)); bigbuf_arcbufs = umem_zalloc(2 * s * sizeof (arc_buf_t *), UMEM_NOFAIL); /* * Iteration 0 test zcopy for DB_UNCACHED dbufs. * Iteration 1 test zcopy to already referenced dbufs. * Iteration 2 test zcopy to dirty dbuf in the same txg. * Iteration 3 test zcopy to dbuf dirty in previous txg. * Iteration 4 test zcopy when dbuf is no longer dirty. * Iteration 5 test zcopy when it can't be done. * Iteration 6 one more zcopy write. */ for (i = 0; i < 7; i++) { uint64_t j; uint64_t off; /* * In iteration 5 (i == 5) use arcbufs * that don't match bigobj blksz to test * dmu_assign_arcbuf_by_dbuf() when it can't directly * assign an arcbuf to a dbuf. */ for (j = 0; j < s; j++) { if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { bigbuf_arcbufs[j] = dmu_request_arcbuf(bonus_db, chunksize); } else { bigbuf_arcbufs[2 * j] = dmu_request_arcbuf(bonus_db, chunksize / 2); bigbuf_arcbufs[2 * j + 1] = dmu_request_arcbuf(bonus_db, chunksize / 2); } } /* * Get a tx for the mods to both packobj and bigobj. */ tx = dmu_tx_create(os); dmu_tx_hold_write(tx, packobj, packoff, packsize); dmu_tx_hold_write(tx, bigobj, bigoff, bigsize); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); for (j = 0; j < s; j++) { if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { dmu_return_arcbuf(bigbuf_arcbufs[j]); } else { dmu_return_arcbuf( bigbuf_arcbufs[2 * j]); dmu_return_arcbuf( bigbuf_arcbufs[2 * j + 1]); } } umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *)); umem_free(od, size); dmu_buf_rele(bonus_db, FTAG); return; } /* * 50% of the time don't read objects in the 1st iteration to * test dmu_assign_arcbuf_by_dbuf() for the case when there are * no existing dbufs for the specified offsets. */ if (i != 0 || ztest_random(2) != 0) { error = dmu_read(os, packobj, packoff, packsize, packbuf, DMU_READ_PREFETCH); ASSERT0(error); error = dmu_read(os, bigobj, bigoff, bigsize, bigbuf, DMU_READ_PREFETCH); ASSERT0(error); } compare_and_update_pbbufs(s, packbuf, bigbuf, bigsize, n, chunksize, txg); /* * We've verified all the old bufwads, and made new ones. * Now write them out. */ dmu_write(os, packobj, packoff, packsize, packbuf, tx); if (ztest_opts.zo_verbose >= 7) { (void) printf("writing offset %"PRIx64" size %"PRIx64"" " txg %"PRIx64"\n", bigoff, bigsize, txg); } for (off = bigoff, j = 0; j < s; j++, off += chunksize) { dmu_buf_t *dbt; if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { memcpy(bigbuf_arcbufs[j]->b_data, (caddr_t)bigbuf + (off - bigoff), chunksize); } else { memcpy(bigbuf_arcbufs[2 * j]->b_data, (caddr_t)bigbuf + (off - bigoff), chunksize / 2); memcpy(bigbuf_arcbufs[2 * j + 1]->b_data, (caddr_t)bigbuf + (off - bigoff) + chunksize / 2, chunksize / 2); } if (i == 1) { VERIFY(dmu_buf_hold(os, bigobj, off, FTAG, &dbt, DMU_READ_NO_PREFETCH) == 0); } if (i != 5 || chunksize < (SPA_MINBLOCKSIZE * 2)) { VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db, off, bigbuf_arcbufs[j], tx)); } else { VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db, off, bigbuf_arcbufs[2 * j], tx)); VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db, off + chunksize / 2, bigbuf_arcbufs[2 * j + 1], tx)); } if (i == 1) { dmu_buf_rele(dbt, FTAG); } } dmu_tx_commit(tx); /* * Sanity check the stuff we just wrote. */ { void *packcheck = umem_alloc(packsize, UMEM_NOFAIL); void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL); VERIFY0(dmu_read(os, packobj, packoff, packsize, packcheck, DMU_READ_PREFETCH)); VERIFY0(dmu_read(os, bigobj, bigoff, bigsize, bigcheck, DMU_READ_PREFETCH)); ASSERT0(memcmp(packbuf, packcheck, packsize)); ASSERT0(memcmp(bigbuf, bigcheck, bigsize)); umem_free(packcheck, packsize); umem_free(bigcheck, bigsize); } if (i == 2) { txg_wait_open(dmu_objset_pool(os), 0, B_TRUE); } else if (i == 3) { txg_wait_synced(dmu_objset_pool(os), 0); } } dmu_buf_rele(bonus_db, FTAG); umem_free(packbuf, packsize); umem_free(bigbuf, bigsize); umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *)); umem_free(od, size); } void ztest_dmu_write_parallel(ztest_ds_t *zd, uint64_t id) { (void) id; ztest_od_t *od; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); uint64_t offset = (1ULL << (ztest_random(20) + 43)) + (ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); /* * Have multiple threads write to large offsets in an object * to verify that parallel writes to an object -- even to the * same blocks within the object -- doesn't cause any trouble. */ ztest_od_init(od, ID_PARALLEL, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) return; while (ztest_random(10) != 0) ztest_io(zd, od->od_object, offset); umem_free(od, sizeof (ztest_od_t)); } void ztest_dmu_prealloc(ztest_ds_t *zd, uint64_t id) { ztest_od_t *od; uint64_t offset = (1ULL << (ztest_random(4) + SPA_MAXBLOCKSHIFT)) + (ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); uint64_t count = ztest_random(20) + 1; uint64_t blocksize = ztest_random_blocksize(); void *data; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } if (ztest_truncate(zd, od->od_object, offset, count * blocksize) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } ztest_prealloc(zd, od->od_object, offset, count * blocksize); data = umem_zalloc(blocksize, UMEM_NOFAIL); while (ztest_random(count) != 0) { uint64_t randoff = offset + (ztest_random(count) * blocksize); if (ztest_write(zd, od->od_object, randoff, blocksize, data) != 0) break; while (ztest_random(4) != 0) ztest_io(zd, od->od_object, randoff); } umem_free(data, blocksize); umem_free(od, sizeof (ztest_od_t)); } /* * Verify that zap_{create,destroy,add,remove,update} work as expected. */ #define ZTEST_ZAP_MIN_INTS 1 #define ZTEST_ZAP_MAX_INTS 4 #define ZTEST_ZAP_MAX_PROPS 1000 void ztest_zap(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t object; uint64_t txg, last_txg; uint64_t value[ZTEST_ZAP_MAX_INTS]; uint64_t zl_ints, zl_intsize, prop; int i, ints; dmu_tx_t *tx; char propname[100], txgname[100]; int error; const char *const hc[2] = { "s.acl.h", ".s.open.h.hyLZlg" }; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) goto out; object = od->od_object; /* * Generate a known hash collision, and verify that * we can lookup and remove both entries. */ tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; for (i = 0; i < 2; i++) { value[i] = i; VERIFY0(zap_add(os, object, hc[i], sizeof (uint64_t), 1, &value[i], tx)); } for (i = 0; i < 2; i++) { VERIFY3U(EEXIST, ==, zap_add(os, object, hc[i], sizeof (uint64_t), 1, &value[i], tx)); VERIFY0( zap_length(os, object, hc[i], &zl_intsize, &zl_ints)); ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, 1); } for (i = 0; i < 2; i++) { VERIFY0(zap_remove(os, object, hc[i], tx)); } dmu_tx_commit(tx); /* * Generate a bunch of random entries. */ ints = MAX(ZTEST_ZAP_MIN_INTS, object % ZTEST_ZAP_MAX_INTS); prop = ztest_random(ZTEST_ZAP_MAX_PROPS); (void) sprintf(propname, "prop_%"PRIu64"", prop); (void) sprintf(txgname, "txg_%"PRIu64"", prop); memset(value, 0, sizeof (value)); last_txg = 0; /* * If these zap entries already exist, validate their contents. */ error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); if (error == 0) { ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, 1); VERIFY0(zap_lookup(os, object, txgname, zl_intsize, zl_ints, &last_txg)); VERIFY0(zap_length(os, object, propname, &zl_intsize, &zl_ints)); ASSERT3U(zl_intsize, ==, sizeof (uint64_t)); ASSERT3U(zl_ints, ==, ints); VERIFY0(zap_lookup(os, object, propname, zl_intsize, zl_ints, value)); for (i = 0; i < ints; i++) { ASSERT3U(value[i], ==, last_txg + object + i); } } else { ASSERT3U(error, ==, ENOENT); } /* * Atomically update two entries in our zap object. * The first is named txg_%llu, and contains the txg * in which the property was last updated. The second * is named prop_%llu, and the nth element of its value * should be txg + object + n. */ tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; if (last_txg > txg) fatal(B_FALSE, "zap future leak: old %"PRIu64" new %"PRIu64"", last_txg, txg); for (i = 0; i < ints; i++) value[i] = txg + object + i; VERIFY0(zap_update(os, object, txgname, sizeof (uint64_t), 1, &txg, tx)); VERIFY0(zap_update(os, object, propname, sizeof (uint64_t), ints, value, tx)); dmu_tx_commit(tx); /* * Remove a random pair of entries. */ prop = ztest_random(ZTEST_ZAP_MAX_PROPS); (void) sprintf(propname, "prop_%"PRIu64"", prop); (void) sprintf(txgname, "txg_%"PRIu64"", prop); error = zap_length(os, object, txgname, &zl_intsize, &zl_ints); if (error == ENOENT) goto out; ASSERT0(error); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; VERIFY0(zap_remove(os, object, txgname, tx)); VERIFY0(zap_remove(os, object, propname, tx)); dmu_tx_commit(tx); out: umem_free(od, sizeof (ztest_od_t)); } /* * Test case to test the upgrading of a microzap to fatzap. */ void ztest_fzap(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t object, txg, value; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), !ztest_random(2)) != 0) goto out; object = od->od_object; /* * Add entries to this ZAP and make sure it spills over * and gets upgraded to a fatzap. Also, since we are adding * 2050 entries we should see ptrtbl growth and leaf-block split. */ for (value = 0; value < 2050; value++) { char name[ZFS_MAX_DATASET_NAME_LEN]; dmu_tx_t *tx; int error; (void) snprintf(name, sizeof (name), "fzap-%"PRIu64"-%"PRIu64"", id, value); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, name); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) goto out; error = zap_add(os, object, name, sizeof (uint64_t), 1, &value, tx); ASSERT(error == 0 || error == EEXIST); dmu_tx_commit(tx); } out: umem_free(od, sizeof (ztest_od_t)); } void ztest_zap_parallel(ztest_ds_t *zd, uint64_t id) { (void) id; objset_t *os = zd->zd_os; ztest_od_t *od; uint64_t txg, object, count, wsize, wc, zl_wsize, zl_wc; dmu_tx_t *tx; int i, namelen, error; int micro = ztest_random(2); char name[20], string_value[20]; void *data; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, ID_PARALLEL, FTAG, micro, DMU_OT_ZAP_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } object = od->od_object; /* * Generate a random name of the form 'xxx.....' where each * x is a random printable character and the dots are dots. * There are 94 such characters, and the name length goes from * 6 to 20, so there are 94^3 * 15 = 12,458,760 possible names. */ namelen = ztest_random(sizeof (name) - 5) + 5 + 1; for (i = 0; i < 3; i++) name[i] = '!' + ztest_random('~' - '!' + 1); for (; i < namelen - 1; i++) name[i] = '.'; name[i] = '\0'; if ((namelen & 1) || micro) { wsize = sizeof (txg); wc = 1; data = &txg; } else { wsize = 1; wc = namelen; data = string_value; } count = -1ULL; VERIFY0(zap_count(os, object, &count)); ASSERT3S(count, !=, -1ULL); /* * Select an operation: length, lookup, add, update, remove. */ i = ztest_random(5); if (i >= 2) { tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, object, B_TRUE, NULL); txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG); if (txg == 0) { umem_free(od, sizeof (ztest_od_t)); return; } memcpy(string_value, name, namelen); } else { tx = NULL; txg = 0; memset(string_value, 0, namelen); } switch (i) { case 0: error = zap_length(os, object, name, &zl_wsize, &zl_wc); if (error == 0) { ASSERT3U(wsize, ==, zl_wsize); ASSERT3U(wc, ==, zl_wc); } else { ASSERT3U(error, ==, ENOENT); } break; case 1: error = zap_lookup(os, object, name, wsize, wc, data); if (error == 0) { if (data == string_value && memcmp(name, data, namelen) != 0) fatal(B_FALSE, "name '%s' != val '%s' len %d", name, (char *)data, namelen); } else { ASSERT3U(error, ==, ENOENT); } break; case 2: error = zap_add(os, object, name, wsize, wc, data, tx); ASSERT(error == 0 || error == EEXIST); break; case 3: VERIFY0(zap_update(os, object, name, wsize, wc, data, tx)); break; case 4: error = zap_remove(os, object, name, tx); ASSERT(error == 0 || error == ENOENT); break; } if (tx != NULL) dmu_tx_commit(tx); umem_free(od, sizeof (ztest_od_t)); } /* * Commit callback data. */ typedef struct ztest_cb_data { list_node_t zcd_node; uint64_t zcd_txg; int zcd_expected_err; boolean_t zcd_added; boolean_t zcd_called; spa_t *zcd_spa; } ztest_cb_data_t; /* This is the actual commit callback function */ static void ztest_commit_callback(void *arg, int error) { ztest_cb_data_t *data = arg; uint64_t synced_txg; VERIFY3P(data, !=, NULL); VERIFY3S(data->zcd_expected_err, ==, error); VERIFY(!data->zcd_called); synced_txg = spa_last_synced_txg(data->zcd_spa); if (data->zcd_txg > synced_txg) fatal(B_FALSE, "commit callback of txg %"PRIu64" called prematurely, " "last synced txg = %"PRIu64"\n", data->zcd_txg, synced_txg); data->zcd_called = B_TRUE; if (error == ECANCELED) { ASSERT0(data->zcd_txg); ASSERT(!data->zcd_added); /* * The private callback data should be destroyed here, but * since we are going to check the zcd_called field after * dmu_tx_abort(), we will destroy it there. */ return; } ASSERT(data->zcd_added); ASSERT3U(data->zcd_txg, !=, 0); (void) mutex_enter(&zcl.zcl_callbacks_lock); /* See if this cb was called more quickly */ if ((synced_txg - data->zcd_txg) < zc_min_txg_delay) zc_min_txg_delay = synced_txg - data->zcd_txg; /* Remove our callback from the list */ list_remove(&zcl.zcl_callbacks, data); (void) mutex_exit(&zcl.zcl_callbacks_lock); umem_free(data, sizeof (ztest_cb_data_t)); } /* Allocate and initialize callback data structure */ static ztest_cb_data_t * ztest_create_cb_data(objset_t *os, uint64_t txg) { ztest_cb_data_t *cb_data; cb_data = umem_zalloc(sizeof (ztest_cb_data_t), UMEM_NOFAIL); cb_data->zcd_txg = txg; cb_data->zcd_spa = dmu_objset_spa(os); list_link_init(&cb_data->zcd_node); return (cb_data); } /* * Commit callback test. */ void ztest_dmu_commit_callbacks(ztest_ds_t *zd, uint64_t id) { objset_t *os = zd->zd_os; ztest_od_t *od; dmu_tx_t *tx; ztest_cb_data_t *cb_data[3], *tmp_cb; uint64_t old_txg, txg; int i, error = 0; od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL); ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) { umem_free(od, sizeof (ztest_od_t)); return; } tx = dmu_tx_create(os); cb_data[0] = ztest_create_cb_data(os, 0); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[0]); dmu_tx_hold_write(tx, od->od_object, 0, sizeof (uint64_t)); /* Every once in a while, abort the transaction on purpose */ if (ztest_random(100) == 0) error = -1; if (!error) error = dmu_tx_assign(tx, TXG_NOWAIT); txg = error ? 0 : dmu_tx_get_txg(tx); cb_data[0]->zcd_txg = txg; cb_data[1] = ztest_create_cb_data(os, txg); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[1]); if (error) { /* * It's not a strict requirement to call the registered * callbacks from inside dmu_tx_abort(), but that's what * it's supposed to happen in the current implementation * so we will check for that. */ for (i = 0; i < 2; i++) { cb_data[i]->zcd_expected_err = ECANCELED; VERIFY(!cb_data[i]->zcd_called); } dmu_tx_abort(tx); for (i = 0; i < 2; i++) { VERIFY(cb_data[i]->zcd_called); umem_free(cb_data[i], sizeof (ztest_cb_data_t)); } umem_free(od, sizeof (ztest_od_t)); return; } cb_data[2] = ztest_create_cb_data(os, txg); dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[2]); /* * Read existing data to make sure there isn't a future leak. */ VERIFY0(dmu_read(os, od->od_object, 0, sizeof (uint64_t), &old_txg, DMU_READ_PREFETCH)); if (old_txg > txg) fatal(B_FALSE, "future leak: got %"PRIu64", open txg is %"PRIu64"", old_txg, txg); dmu_write(os, od->od_object, 0, sizeof (uint64_t), &txg, tx); (void) mutex_enter(&zcl.zcl_callbacks_lock); /* * Since commit callbacks don't have any ordering requirement and since * it is theoretically possible for a commit callback to be called * after an arbitrary amount of time has elapsed since its txg has been * synced, it is difficult to reliably determine whether a commit * callback hasn't been called due to high load or due to a flawed * implementation. * * In practice, we will assume that if after a certain number of txgs a * commit callback hasn't been called, then most likely there's an * implementation bug.. */ tmp_cb = list_head(&zcl.zcl_callbacks); if (tmp_cb != NULL && tmp_cb->zcd_txg + ZTEST_COMMIT_CB_THRESH < txg) { fatal(B_FALSE, "Commit callback threshold exceeded, " "oldest txg: %"PRIu64", open txg: %"PRIu64"\n", tmp_cb->zcd_txg, txg); } /* * Let's find the place to insert our callbacks. * * Even though the list is ordered by txg, it is possible for the * insertion point to not be the end because our txg may already be * quiescing at this point and other callbacks in the open txg * (from other objsets) may have sneaked in. */ tmp_cb = list_tail(&zcl.zcl_callbacks); while (tmp_cb != NULL && tmp_cb->zcd_txg > txg) tmp_cb = list_prev(&zcl.zcl_callbacks, tmp_cb); /* Add the 3 callbacks to the list */ for (i = 0; i < 3; i++) { if (tmp_cb == NULL) list_insert_head(&zcl.zcl_callbacks, cb_data[i]); else list_insert_after(&zcl.zcl_callbacks, tmp_cb, cb_data[i]); cb_data[i]->zcd_added = B_TRUE; VERIFY(!cb_data[i]->zcd_called); tmp_cb = cb_data[i]; } zc_cb_counter += 3; (void) mutex_exit(&zcl.zcl_callbacks_lock); dmu_tx_commit(tx); umem_free(od, sizeof (ztest_od_t)); } /* * Visit each object in the dataset. Verify that its properties * are consistent what was stored in the block tag when it was created, * and that its unused bonus buffer space has not been overwritten. */ void ztest_verify_dnode_bt(ztest_ds_t *zd, uint64_t id) { (void) id; objset_t *os = zd->zd_os; uint64_t obj; int err = 0; for (obj = 0; err == 0; err = dmu_object_next(os, &obj, FALSE, 0)) { ztest_block_tag_t *bt = NULL; dmu_object_info_t doi; dmu_buf_t *db; ztest_object_lock(zd, obj, RL_READER); if (dmu_bonus_hold(os, obj, FTAG, &db) != 0) { ztest_object_unlock(zd, obj); continue; } dmu_object_info_from_db(db, &doi); if (doi.doi_bonus_size >= sizeof (*bt)) bt = ztest_bt_bonus(db); if (bt && bt->bt_magic == BT_MAGIC) { ztest_bt_verify(bt, os, obj, doi.doi_dnodesize, bt->bt_offset, bt->bt_gen, bt->bt_txg, bt->bt_crtxg); ztest_verify_unused_bonus(db, bt, obj, os, bt->bt_gen); } dmu_buf_rele(db, FTAG); ztest_object_unlock(zd, obj); } } void ztest_dsl_prop_get_set(ztest_ds_t *zd, uint64_t id) { (void) id; zfs_prop_t proplist[] = { ZFS_PROP_CHECKSUM, ZFS_PROP_COMPRESSION, ZFS_PROP_COPIES, ZFS_PROP_DEDUP }; (void) pthread_rwlock_rdlock(&ztest_name_lock); for (int p = 0; p < sizeof (proplist) / sizeof (proplist[0]); p++) (void) ztest_dsl_prop_set_uint64(zd->zd_name, proplist[p], ztest_random_dsl_prop(proplist[p]), (int)ztest_random(2)); VERIFY0(ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_RECORDSIZE, ztest_random_blocksize(), (int)ztest_random(2))); (void) pthread_rwlock_unlock(&ztest_name_lock); } void ztest_spa_prop_get_set(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; nvlist_t *props = NULL; (void) pthread_rwlock_rdlock(&ztest_name_lock); (void) ztest_spa_prop_set_uint64(ZPOOL_PROP_AUTOTRIM, ztest_random(2)); VERIFY0(spa_prop_get(ztest_spa, &props)); if (ztest_opts.zo_verbose >= 6) dump_nvlist(props, 4); fnvlist_free(props); (void) pthread_rwlock_unlock(&ztest_name_lock); } static int user_release_one(const char *snapname, const char *holdname) { nvlist_t *snaps, *holds; int error; snaps = fnvlist_alloc(); holds = fnvlist_alloc(); fnvlist_add_boolean(holds, holdname); fnvlist_add_nvlist(snaps, snapname, holds); fnvlist_free(holds); error = dsl_dataset_user_release(snaps, NULL); fnvlist_free(snaps); return (error); } /* * Test snapshot hold/release and deferred destroy. */ void ztest_dmu_snapshot_hold(ztest_ds_t *zd, uint64_t id) { int error; objset_t *os = zd->zd_os; objset_t *origin; char snapname[100]; char fullname[100]; char clonename[100]; char tag[100]; char osname[ZFS_MAX_DATASET_NAME_LEN]; nvlist_t *holds; (void) pthread_rwlock_rdlock(&ztest_name_lock); dmu_objset_name(os, osname); (void) snprintf(snapname, sizeof (snapname), "sh1_%"PRIu64"", id); (void) snprintf(fullname, sizeof (fullname), "%s@%s", osname, snapname); (void) snprintf(clonename, sizeof (clonename), "%s/ch1_%"PRIu64"", osname, id); (void) snprintf(tag, sizeof (tag), "tag_%"PRIu64"", id); /* * Clean up from any previous run. */ error = dsl_destroy_head(clonename); if (error != ENOENT) ASSERT0(error); error = user_release_one(fullname, tag); if (error != ESRCH && error != ENOENT) ASSERT0(error); error = dsl_destroy_snapshot(fullname, B_FALSE); if (error != ENOENT) ASSERT0(error); /* * Create snapshot, clone it, mark snap for deferred destroy, * destroy clone, verify snap was also destroyed. */ error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); goto out; } fatal(B_FALSE, "dmu_objset_snapshot(%s) = %d", fullname, error); } error = dmu_objset_clone(clonename, fullname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_clone"); goto out; } fatal(B_FALSE, "dmu_objset_clone(%s) = %d", clonename, error); } error = dsl_destroy_snapshot(fullname, B_TRUE); if (error) { fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_TRUE) = %d", fullname, error); } error = dsl_destroy_head(clonename); if (error) fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clonename, error); error = dmu_objset_hold(fullname, FTAG, &origin); if (error != ENOENT) fatal(B_FALSE, "dmu_objset_hold(%s) = %d", fullname, error); /* * Create snapshot, add temporary hold, verify that we can't * destroy a held snapshot, mark for deferred destroy, * release hold, verify snapshot was destroyed. */ error = dmu_objset_snapshot_one(osname, snapname); if (error) { if (error == ENOSPC) { ztest_record_enospc("dmu_objset_snapshot"); goto out; } fatal(B_FALSE, "dmu_objset_snapshot(%s) = %d", fullname, error); } holds = fnvlist_alloc(); fnvlist_add_string(holds, fullname, tag); error = dsl_dataset_user_hold(holds, 0, NULL); fnvlist_free(holds); if (error == ENOSPC) { ztest_record_enospc("dsl_dataset_user_hold"); goto out; } else if (error) { fatal(B_FALSE, "dsl_dataset_user_hold(%s, %s) = %u", fullname, tag, error); } error = dsl_destroy_snapshot(fullname, B_FALSE); if (error != EBUSY) { fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_FALSE) = %d", fullname, error); } error = dsl_destroy_snapshot(fullname, B_TRUE); if (error) { fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_TRUE) = %d", fullname, error); } error = user_release_one(fullname, tag); if (error) fatal(B_FALSE, "user_release_one(%s, %s) = %d", fullname, tag, error); VERIFY3U(dmu_objset_hold(fullname, FTAG, &origin), ==, ENOENT); out: (void) pthread_rwlock_unlock(&ztest_name_lock); } /* * Inject random faults into the on-disk data. */ void ztest_fault_inject(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; ztest_shared_t *zs = ztest_shared; spa_t *spa = ztest_spa; int fd; uint64_t offset; uint64_t leaves; uint64_t bad = 0x1990c0ffeedecadeull; uint64_t top, leaf; char *path0; char *pathrand; size_t fsize; int bshift = SPA_MAXBLOCKSHIFT + 2; int iters = 1000; int maxfaults; int mirror_save; vdev_t *vd0 = NULL; uint64_t guid0 = 0; boolean_t islog = B_FALSE; path0 = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); pathrand = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); mutex_enter(&ztest_vdev_lock); /* * Device removal is in progress, fault injection must be disabled * until it completes and the pool is scrubbed. The fault injection * strategy for damaging blocks does not take in to account evacuated * blocks which may have already been damaged. */ if (ztest_device_removal_active) { mutex_exit(&ztest_vdev_lock); goto out; } maxfaults = MAXFAULTS(zs); leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raid_children; mirror_save = zs->zs_mirrors; mutex_exit(&ztest_vdev_lock); ASSERT3U(leaves, >=, 1); /* * While ztest is running the number of leaves will not change. This * is critical for the fault injection logic as it determines where * errors can be safely injected such that they are always repairable. * * When restarting ztest a different number of leaves may be requested * which will shift the regions to be damaged. This is fine as long * as the pool has been scrubbed prior to using the new mapping. * Failure to do can result in non-repairable damage being injected. */ if (ztest_pool_scrubbed == B_FALSE) goto out; /* * Grab the name lock as reader. There are some operations * which don't like to have their vdevs changed while * they are in progress (i.e. spa_change_guid). Those * operations will have grabbed the name lock as writer. */ (void) pthread_rwlock_rdlock(&ztest_name_lock); /* * We need SCL_STATE here because we're going to look at vd0->vdev_tsd. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); if (ztest_random(2) == 0) { /* * Inject errors on a normal data device or slog device. */ top = ztest_random_vdev_top(spa, B_TRUE); leaf = ztest_random(leaves) + zs->zs_splits; /* * Generate paths to the first leaf in this top-level vdev, * and to the random leaf we selected. We'll induce transient * write failures and random online/offline activity on leaf 0, * and we'll write random garbage to the randomly chosen leaf. */ (void) snprintf(path0, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + zs->zs_splits); (void) snprintf(pathrand, MAXPATHLEN, ztest_dev_template, ztest_opts.zo_dir, ztest_opts.zo_pool, top * leaves + leaf); vd0 = vdev_lookup_by_path(spa->spa_root_vdev, path0); if (vd0 != NULL && vd0->vdev_top->vdev_islog) islog = B_TRUE; /* * If the top-level vdev needs to be resilvered * then we only allow faults on the device that is * resilvering. */ if (vd0 != NULL && maxfaults != 1 && (!vdev_resilver_needed(vd0->vdev_top, NULL, NULL) || vd0->vdev_resilver_txg != 0)) { /* * Make vd0 explicitly claim to be unreadable, * or unwritable, or reach behind its back * and close the underlying fd. We can do this if * maxfaults == 0 because we'll fail and reexecute, * and we can do it if maxfaults >= 2 because we'll * have enough redundancy. If maxfaults == 1, the * combination of this with injection of random data * corruption below exceeds the pool's fault tolerance. */ vdev_file_t *vf = vd0->vdev_tsd; zfs_dbgmsg("injecting fault to vdev %llu; maxfaults=%d", (long long)vd0->vdev_id, (int)maxfaults); if (vf != NULL && ztest_random(3) == 0) { (void) close(vf->vf_file->f_fd); vf->vf_file->f_fd = -1; } else if (ztest_random(2) == 0) { vd0->vdev_cant_read = B_TRUE; } else { vd0->vdev_cant_write = B_TRUE; } guid0 = vd0->vdev_guid; } } else { /* * Inject errors on an l2cache device. */ spa_aux_vdev_t *sav = &spa->spa_l2cache; if (sav->sav_count == 0) { spa_config_exit(spa, SCL_STATE, FTAG); (void) pthread_rwlock_unlock(&ztest_name_lock); goto out; } vd0 = sav->sav_vdevs[ztest_random(sav->sav_count)]; guid0 = vd0->vdev_guid; (void) strlcpy(path0, vd0->vdev_path, MAXPATHLEN); (void) strlcpy(pathrand, vd0->vdev_path, MAXPATHLEN); leaf = 0; leaves = 1; maxfaults = INT_MAX; /* no limit on cache devices */ } spa_config_exit(spa, SCL_STATE, FTAG); (void) pthread_rwlock_unlock(&ztest_name_lock); /* * If we can tolerate two or more faults, or we're dealing * with a slog, randomly online/offline vd0. */ if ((maxfaults >= 2 || islog) && guid0 != 0) { if (ztest_random(10) < 6) { int flags = (ztest_random(2) == 0 ? ZFS_OFFLINE_TEMPORARY : 0); /* * We have to grab the zs_name_lock as writer to * prevent a race between offlining a slog and * destroying a dataset. Offlining the slog will * grab a reference on the dataset which may cause * dsl_destroy_head() to fail with EBUSY thus * leaving the dataset in an inconsistent state. */ if (islog) (void) pthread_rwlock_wrlock(&ztest_name_lock); VERIFY3U(vdev_offline(spa, guid0, flags), !=, EBUSY); if (islog) (void) pthread_rwlock_unlock(&ztest_name_lock); } else { /* * Ideally we would like to be able to randomly * call vdev_[on|off]line without holding locks * to force unpredictable failures but the side * effects of vdev_[on|off]line prevent us from * doing so. We grab the ztest_vdev_lock here to * prevent a race between injection testing and * aux_vdev removal. */ mutex_enter(&ztest_vdev_lock); (void) vdev_online(spa, guid0, 0, NULL); mutex_exit(&ztest_vdev_lock); } } if (maxfaults == 0) goto out; /* * We have at least single-fault tolerance, so inject data corruption. */ fd = open(pathrand, O_RDWR); if (fd == -1) /* we hit a gap in the device namespace */ goto out; fsize = lseek(fd, 0, SEEK_END); while (--iters != 0) { /* * The offset must be chosen carefully to ensure that * we do not inject a given logical block with errors * on two different leaf devices, because ZFS can not * tolerate that (if maxfaults==1). * * To achieve this we divide each leaf device into * chunks of size (# leaves * SPA_MAXBLOCKSIZE * 4). * Each chunk is further divided into error-injection * ranges (can accept errors) and clear ranges (we do * not inject errors in those). Each error-injection * range can accept errors only for a single leaf vdev. * Error-injection ranges are separated by clear ranges. * * For example, with 3 leaves, each chunk looks like: * 0 to 32M: injection range for leaf 0 * 32M to 64M: clear range - no injection allowed * 64M to 96M: injection range for leaf 1 * 96M to 128M: clear range - no injection allowed * 128M to 160M: injection range for leaf 2 * 160M to 192M: clear range - no injection allowed * * Each clear range must be large enough such that a * single block cannot straddle it. This way a block * can't be a target in two different injection ranges * (on different leaf vdevs). */ offset = ztest_random(fsize / (leaves << bshift)) * (leaves << bshift) + (leaf << bshift) + (ztest_random(1ULL << (bshift - 1)) & -8ULL); /* * Only allow damage to the labels at one end of the vdev. * * If all labels are damaged, the device will be totally * inaccessible, which will result in loss of data, * because we also damage (parts of) the other side of * the mirror/raidz. * * Additionally, we will always have both an even and an * odd label, so that we can handle crashes in the * middle of vdev_config_sync(). */ if ((leaf & 1) == 0 && offset < VDEV_LABEL_START_SIZE) continue; /* * The two end labels are stored at the "end" of the disk, but * the end of the disk (vdev_psize) is aligned to * sizeof (vdev_label_t). */ uint64_t psize = P2ALIGN(fsize, sizeof (vdev_label_t)); if ((leaf & 1) == 1 && offset + sizeof (bad) > psize - VDEV_LABEL_END_SIZE) continue; mutex_enter(&ztest_vdev_lock); if (mirror_save != zs->zs_mirrors) { mutex_exit(&ztest_vdev_lock); (void) close(fd); goto out; } if (pwrite(fd, &bad, sizeof (bad), offset) != sizeof (bad)) fatal(B_TRUE, "can't inject bad word at 0x%"PRIx64" in %s", offset, pathrand); mutex_exit(&ztest_vdev_lock); if (ztest_opts.zo_verbose >= 7) (void) printf("injected bad word into %s," " offset 0x%"PRIx64"\n", pathrand, offset); } (void) close(fd); out: umem_free(path0, MAXPATHLEN); umem_free(pathrand, MAXPATHLEN); } /* * By design ztest will never inject uncorrectable damage in to the pool. * Issue a scrub, wait for it to complete, and verify there is never any * persistent damage. * * Only after a full scrub has been completed is it safe to start injecting * data corruption. See the comment in zfs_fault_inject(). */ static int ztest_scrub_impl(spa_t *spa) { int error = spa_scan(spa, POOL_SCAN_SCRUB); if (error) return (error); while (dsl_scan_scrubbing(spa_get_dsl(spa))) txg_wait_synced(spa_get_dsl(spa), 0); if (spa_get_errlog_size(spa) > 0) return (ECKSUM); ztest_pool_scrubbed = B_TRUE; return (0); } /* * Scrub the pool. */ void ztest_scrub(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; int error; /* * Scrub in progress by device removal. */ if (ztest_device_removal_active) return; /* * Start a scrub, wait a moment, then force a restart. */ (void) spa_scan(spa, POOL_SCAN_SCRUB); (void) poll(NULL, 0, 100); error = ztest_scrub_impl(spa); if (error == EBUSY) error = 0; ASSERT0(error); } /* * Change the guid for the pool. */ void ztest_reguid(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; uint64_t orig, load; int error; if (ztest_opts.zo_mmp_test) return; orig = spa_guid(spa); load = spa_load_guid(spa); (void) pthread_rwlock_wrlock(&ztest_name_lock); error = spa_change_guid(spa); (void) pthread_rwlock_unlock(&ztest_name_lock); if (error != 0) return; if (ztest_opts.zo_verbose >= 4) { (void) printf("Changed guid old %"PRIu64" -> %"PRIu64"\n", orig, spa_guid(spa)); } VERIFY3U(orig, !=, spa_guid(spa)); VERIFY3U(load, ==, spa_load_guid(spa)); } void ztest_blake3(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; hrtime_t end = gethrtime() + NANOSEC; zio_cksum_salt_t salt; void *salt_ptr = &salt.zcs_bytes; struct abd *abd_data, *abd_meta; void *buf, *templ; int i, *ptr; uint32_t size; BLAKE3_CTX ctx; size = ztest_random_blocksize(); buf = umem_alloc(size, UMEM_NOFAIL); abd_data = abd_alloc(size, B_FALSE); abd_meta = abd_alloc(size, B_TRUE); for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++) *ptr = ztest_random(UINT_MAX); memset(salt_ptr, 'A', 32); abd_copy_from_buf_off(abd_data, buf, 0, size); abd_copy_from_buf_off(abd_meta, buf, 0, size); while (gethrtime() <= end) { int run_count = 100; zio_cksum_t zc_ref1, zc_ref2; zio_cksum_t zc_res1, zc_res2; void *ref1 = &zc_ref1; void *ref2 = &zc_ref2; void *res1 = &zc_res1; void *res2 = &zc_res2; /* BLAKE3_KEY_LEN = 32 */ VERIFY0(blake3_impl_setname("generic")); templ = abd_checksum_blake3_tmpl_init(&salt); Blake3_InitKeyed(&ctx, salt_ptr); Blake3_Update(&ctx, buf, size); Blake3_Final(&ctx, ref1); zc_ref2 = zc_ref1; ZIO_CHECKSUM_BSWAP(&zc_ref2); abd_checksum_blake3_tmpl_free(templ); VERIFY0(blake3_impl_setname("cycle")); while (run_count-- > 0) { /* Test current implementation */ Blake3_InitKeyed(&ctx, salt_ptr); Blake3_Update(&ctx, buf, size); Blake3_Final(&ctx, res1); zc_res2 = zc_res1; ZIO_CHECKSUM_BSWAP(&zc_res2); VERIFY0(memcmp(ref1, res1, 32)); VERIFY0(memcmp(ref2, res2, 32)); /* Test ABD - data */ templ = abd_checksum_blake3_tmpl_init(&salt); abd_checksum_blake3_native(abd_data, size, templ, &zc_res1); abd_checksum_blake3_byteswap(abd_data, size, templ, &zc_res2); VERIFY0(memcmp(ref1, res1, 32)); VERIFY0(memcmp(ref2, res2, 32)); /* Test ABD - metadata */ abd_checksum_blake3_native(abd_meta, size, templ, &zc_res1); abd_checksum_blake3_byteswap(abd_meta, size, templ, &zc_res2); abd_checksum_blake3_tmpl_free(templ); VERIFY0(memcmp(ref1, res1, 32)); VERIFY0(memcmp(ref2, res2, 32)); } } abd_free(abd_data); abd_free(abd_meta); umem_free(buf, size); } void ztest_fletcher(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; hrtime_t end = gethrtime() + NANOSEC; while (gethrtime() <= end) { int run_count = 100; void *buf; struct abd *abd_data, *abd_meta; uint32_t size; int *ptr; int i; zio_cksum_t zc_ref; zio_cksum_t zc_ref_byteswap; size = ztest_random_blocksize(); buf = umem_alloc(size, UMEM_NOFAIL); abd_data = abd_alloc(size, B_FALSE); abd_meta = abd_alloc(size, B_TRUE); for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++) *ptr = ztest_random(UINT_MAX); abd_copy_from_buf_off(abd_data, buf, 0, size); abd_copy_from_buf_off(abd_meta, buf, 0, size); VERIFY0(fletcher_4_impl_set("scalar")); fletcher_4_native(buf, size, NULL, &zc_ref); fletcher_4_byteswap(buf, size, NULL, &zc_ref_byteswap); VERIFY0(fletcher_4_impl_set("cycle")); while (run_count-- > 0) { zio_cksum_t zc; zio_cksum_t zc_byteswap; fletcher_4_byteswap(buf, size, NULL, &zc_byteswap); fletcher_4_native(buf, size, NULL, &zc); VERIFY0(memcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(memcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); /* Test ABD - data */ abd_fletcher_4_byteswap(abd_data, size, NULL, &zc_byteswap); abd_fletcher_4_native(abd_data, size, NULL, &zc); VERIFY0(memcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(memcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); /* Test ABD - metadata */ abd_fletcher_4_byteswap(abd_meta, size, NULL, &zc_byteswap); abd_fletcher_4_native(abd_meta, size, NULL, &zc); VERIFY0(memcmp(&zc, &zc_ref, sizeof (zc))); VERIFY0(memcmp(&zc_byteswap, &zc_ref_byteswap, sizeof (zc_byteswap))); } umem_free(buf, size); abd_free(abd_data); abd_free(abd_meta); } } void ztest_fletcher_incr(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; void *buf; size_t size; int *ptr; int i; zio_cksum_t zc_ref; zio_cksum_t zc_ref_bswap; hrtime_t end = gethrtime() + NANOSEC; while (gethrtime() <= end) { int run_count = 100; size = ztest_random_blocksize(); buf = umem_alloc(size, UMEM_NOFAIL); for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++) *ptr = ztest_random(UINT_MAX); VERIFY0(fletcher_4_impl_set("scalar")); fletcher_4_native(buf, size, NULL, &zc_ref); fletcher_4_byteswap(buf, size, NULL, &zc_ref_bswap); VERIFY0(fletcher_4_impl_set("cycle")); while (run_count-- > 0) { zio_cksum_t zc; zio_cksum_t zc_bswap; size_t pos = 0; ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0); ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0); while (pos < size) { size_t inc = 64 * ztest_random(size / 67); /* sometimes add few bytes to test non-simd */ if (ztest_random(100) < 10) inc += P2ALIGN(ztest_random(64), sizeof (uint32_t)); if (inc > (size - pos)) inc = size - pos; fletcher_4_incremental_native(buf + pos, inc, &zc); fletcher_4_incremental_byteswap(buf + pos, inc, &zc_bswap); pos += inc; } VERIFY3U(pos, ==, size); VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref)); VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap)); /* * verify if incremental on the whole buffer is * equivalent to non-incremental version */ ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0); ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0); fletcher_4_incremental_native(buf, size, &zc); fletcher_4_incremental_byteswap(buf, size, &zc_bswap); VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref)); VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap)); } umem_free(buf, size); } } static int ztest_set_global_vars(void) { for (size_t i = 0; i < ztest_opts.zo_gvars_count; i++) { char *kv = ztest_opts.zo_gvars[i]; VERIFY3U(strlen(kv), <=, ZO_GVARS_MAX_ARGLEN); VERIFY3U(strlen(kv), >, 0); int err = set_global_var(kv); if (ztest_opts.zo_verbose > 0) { (void) printf("setting global var %s ... %s\n", kv, err ? "failed" : "ok"); } if (err != 0) { (void) fprintf(stderr, "failed to set global var '%s'\n", kv); return (err); } } return (0); } static char ** ztest_global_vars_to_zdb_args(void) { char **args = calloc(2*ztest_opts.zo_gvars_count + 1, sizeof (char *)); char **cur = args; for (size_t i = 0; i < ztest_opts.zo_gvars_count; i++) { *cur++ = (char *)"-o"; *cur++ = ztest_opts.zo_gvars[i]; } ASSERT3P(cur, ==, &args[2*ztest_opts.zo_gvars_count]); *cur = NULL; return (args); } /* The end of strings is indicated by a NULL element */ static char * join_strings(char **strings, const char *sep) { size_t totallen = 0; for (char **sp = strings; *sp != NULL; sp++) { totallen += strlen(*sp); totallen += strlen(sep); } if (totallen > 0) { ASSERT(totallen >= strlen(sep)); totallen -= strlen(sep); } size_t buflen = totallen + 1; char *o = malloc(buflen); /* trailing 0 byte */ o[0] = '\0'; for (char **sp = strings; *sp != NULL; sp++) { size_t would; would = strlcat(o, *sp, buflen); VERIFY3U(would, <, buflen); if (*(sp+1) == NULL) { break; } would = strlcat(o, sep, buflen); VERIFY3U(would, <, buflen); } ASSERT3S(strlen(o), ==, totallen); return (o); } static int ztest_check_path(char *path) { struct stat s; /* return true on success */ return (!stat(path, &s)); } static void ztest_get_zdb_bin(char *bin, int len) { char *zdb_path; /* * Try to use $ZDB and in-tree zdb path. If not successful, just * let popen to search through PATH. */ if ((zdb_path = getenv("ZDB"))) { strlcpy(bin, zdb_path, len); /* In env */ if (!ztest_check_path(bin)) { ztest_dump_core = 0; fatal(B_TRUE, "invalid ZDB '%s'", bin); } return; } VERIFY3P(realpath(getexecname(), bin), !=, NULL); if (strstr(bin, ".libs/ztest")) { strstr(bin, ".libs/ztest")[0] = '\0'; /* In-tree */ strcat(bin, "zdb"); if (ztest_check_path(bin)) return; } strcpy(bin, "zdb"); } static vdev_t * ztest_random_concrete_vdev_leaf(vdev_t *vd) { if (vd == NULL) return (NULL); if (vd->vdev_children == 0) return (vd); vdev_t *eligible[vd->vdev_children]; int eligible_idx = 0, i; for (i = 0; i < vd->vdev_children; i++) { vdev_t *cvd = vd->vdev_child[i]; if (cvd->vdev_top->vdev_removing) continue; if (cvd->vdev_children > 0 || (vdev_is_concrete(cvd) && !cvd->vdev_detached)) { eligible[eligible_idx++] = cvd; } } VERIFY3S(eligible_idx, >, 0); uint64_t child_no = ztest_random(eligible_idx); return (ztest_random_concrete_vdev_leaf(eligible[child_no])); } void ztest_initialize(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; int error = 0; mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* Random leaf vdev */ vdev_t *rand_vd = ztest_random_concrete_vdev_leaf(spa->spa_root_vdev); if (rand_vd == NULL) { spa_config_exit(spa, SCL_VDEV, FTAG); mutex_exit(&ztest_vdev_lock); return; } /* * The random vdev we've selected may change as soon as we * drop the spa_config_lock. We create local copies of things * we're interested in. */ uint64_t guid = rand_vd->vdev_guid; char *path = strdup(rand_vd->vdev_path); boolean_t active = rand_vd->vdev_initialize_thread != NULL; zfs_dbgmsg("vd %px, guid %llu", rand_vd, (u_longlong_t)guid); spa_config_exit(spa, SCL_VDEV, FTAG); uint64_t cmd = ztest_random(POOL_INITIALIZE_FUNCS); nvlist_t *vdev_guids = fnvlist_alloc(); nvlist_t *vdev_errlist = fnvlist_alloc(); fnvlist_add_uint64(vdev_guids, path, guid); error = spa_vdev_initialize(spa, vdev_guids, cmd, vdev_errlist); fnvlist_free(vdev_guids); fnvlist_free(vdev_errlist); switch (cmd) { case POOL_INITIALIZE_CANCEL: if (ztest_opts.zo_verbose >= 4) { (void) printf("Cancel initialize %s", path); if (!active) (void) printf(" failed (no initialize active)"); (void) printf("\n"); } break; case POOL_INITIALIZE_START: if (ztest_opts.zo_verbose >= 4) { (void) printf("Start initialize %s", path); if (active && error == 0) (void) printf(" failed (already active)"); else if (error != 0) (void) printf(" failed (error %d)", error); (void) printf("\n"); } break; case POOL_INITIALIZE_SUSPEND: if (ztest_opts.zo_verbose >= 4) { (void) printf("Suspend initialize %s", path); if (!active) (void) printf(" failed (no initialize active)"); (void) printf("\n"); } break; } free(path); mutex_exit(&ztest_vdev_lock); } void ztest_trim(ztest_ds_t *zd, uint64_t id) { (void) zd, (void) id; spa_t *spa = ztest_spa; int error = 0; mutex_enter(&ztest_vdev_lock); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* Random leaf vdev */ vdev_t *rand_vd = ztest_random_concrete_vdev_leaf(spa->spa_root_vdev); if (rand_vd == NULL) { spa_config_exit(spa, SCL_VDEV, FTAG); mutex_exit(&ztest_vdev_lock); return; } /* * The random vdev we've selected may change as soon as we * drop the spa_config_lock. We create local copies of things * we're interested in. */ uint64_t guid = rand_vd->vdev_guid; char *path = strdup(rand_vd->vdev_path); boolean_t active = rand_vd->vdev_trim_thread != NULL; zfs_dbgmsg("vd %p, guid %llu", rand_vd, (u_longlong_t)guid); spa_config_exit(spa, SCL_VDEV, FTAG); uint64_t cmd = ztest_random(POOL_TRIM_FUNCS); uint64_t rate = 1 << ztest_random(30); boolean_t partial = (ztest_random(5) > 0); boolean_t secure = (ztest_random(5) > 0); nvlist_t *vdev_guids = fnvlist_alloc(); nvlist_t *vdev_errlist = fnvlist_alloc(); fnvlist_add_uint64(vdev_guids, path, guid); error = spa_vdev_trim(spa, vdev_guids, cmd, rate, partial, secure, vdev_errlist); fnvlist_free(vdev_guids); fnvlist_free(vdev_errlist); switch (cmd) { case POOL_TRIM_CANCEL: if (ztest_opts.zo_verbose >= 4) { (void) printf("Cancel TRIM %s", path); if (!active) (void) printf(" failed (no TRIM active)"); (void) printf("\n"); } break; case POOL_TRIM_START: if (ztest_opts.zo_verbose >= 4) { (void) printf("Start TRIM %s", path); if (active && error == 0) (void) printf(" failed (already active)"); else if (error != 0) (void) printf(" failed (error %d)", error); (void) printf("\n"); } break; case POOL_TRIM_SUSPEND: if (ztest_opts.zo_verbose >= 4) { (void) printf("Suspend TRIM %s", path); if (!active) (void) printf(" failed (no TRIM active)"); (void) printf("\n"); } break; } free(path); mutex_exit(&ztest_vdev_lock); } /* * Verify pool integrity by running zdb. */ static void ztest_run_zdb(const char *pool) { int status; char *bin; char *zdb; char *zbuf; const int len = MAXPATHLEN + MAXNAMELEN + 20; FILE *fp; bin = umem_alloc(len, UMEM_NOFAIL); zdb = umem_alloc(len, UMEM_NOFAIL); zbuf = umem_alloc(1024, UMEM_NOFAIL); ztest_get_zdb_bin(bin, len); char **set_gvars_args = ztest_global_vars_to_zdb_args(); char *set_gvars_args_joined = join_strings(set_gvars_args, " "); free(set_gvars_args); size_t would = snprintf(zdb, len, "%s -bcc%s%s -G -d -Y -e -y %s -p %s %s", bin, ztest_opts.zo_verbose >= 3 ? "s" : "", ztest_opts.zo_verbose >= 4 ? "v" : "", set_gvars_args_joined, ztest_opts.zo_dir, pool); ASSERT3U(would, <, len); free(set_gvars_args_joined); if (ztest_opts.zo_verbose >= 5) (void) printf("Executing %s\n", zdb); fp = popen(zdb, "r"); while (fgets(zbuf, 1024, fp) != NULL) if (ztest_opts.zo_verbose >= 3) (void) printf("%s", zbuf); status = pclose(fp); if (status == 0) goto out; ztest_dump_core = 0; if (WIFEXITED(status)) fatal(B_FALSE, "'%s' exit code %d", zdb, WEXITSTATUS(status)); else fatal(B_FALSE, "'%s' died with signal %d", zdb, WTERMSIG(status)); out: umem_free(bin, len); umem_free(zdb, len); umem_free(zbuf, 1024); } static void ztest_walk_pool_directory(const char *header) { spa_t *spa = NULL; if (ztest_opts.zo_verbose >= 6) (void) puts(header); mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) if (ztest_opts.zo_verbose >= 6) (void) printf("\t%s\n", spa_name(spa)); mutex_exit(&spa_namespace_lock); } static void ztest_spa_import_export(char *oldname, char *newname) { nvlist_t *config, *newconfig; uint64_t pool_guid; spa_t *spa; int error; if (ztest_opts.zo_verbose >= 4) { (void) printf("import/export: old = %s, new = %s\n", oldname, newname); } /* * Clean up from previous runs. */ (void) spa_destroy(newname); /* * Get the pool's configuration and guid. */ VERIFY0(spa_open(oldname, &spa, FTAG)); /* * Kick off a scrub to tickle scrub/export races. */ if (ztest_random(2) == 0) (void) spa_scan(spa, POOL_SCAN_SCRUB); pool_guid = spa_guid(spa); spa_close(spa, FTAG); ztest_walk_pool_directory("pools before export"); /* * Export it. */ VERIFY0(spa_export(oldname, &config, B_FALSE, B_FALSE)); ztest_walk_pool_directory("pools after export"); /* * Try to import it. */ newconfig = spa_tryimport(config); ASSERT3P(newconfig, !=, NULL); fnvlist_free(newconfig); /* * Import it under the new name. */ error = spa_import(newname, config, NULL, 0); if (error != 0) { dump_nvlist(config, 0); fatal(B_FALSE, "couldn't import pool %s as %s: error %u", oldname, newname, error); } ztest_walk_pool_directory("pools after import"); /* * Try to import it again -- should fail with EEXIST. */ VERIFY3U(EEXIST, ==, spa_import(newname, config, NULL, 0)); /* * Try to import it under a different name -- should fail with EEXIST. */ VERIFY3U(EEXIST, ==, spa_import(oldname, config, NULL, 0)); /* * Verify that the pool is no longer visible under the old name. */ VERIFY3U(ENOENT, ==, spa_open(oldname, &spa, FTAG)); /* * Verify that we can open and close the pool using the new name. */ VERIFY0(spa_open(newname, &spa, FTAG)); ASSERT3U(pool_guid, ==, spa_guid(spa)); spa_close(spa, FTAG); fnvlist_free(config); } static void ztest_resume(spa_t *spa) { if (spa_suspended(spa) && ztest_opts.zo_verbose >= 6) (void) printf("resuming from suspended state\n"); spa_vdev_state_enter(spa, SCL_NONE); vdev_clear(spa, NULL); (void) spa_vdev_state_exit(spa, NULL, 0); (void) zio_resume(spa); } static __attribute__((noreturn)) void ztest_resume_thread(void *arg) { spa_t *spa = arg; while (!ztest_exiting) { if (spa_suspended(spa)) ztest_resume(spa); (void) poll(NULL, 0, 100); /* * Periodically change the zfs_compressed_arc_enabled setting. */ if (ztest_random(10) == 0) zfs_compressed_arc_enabled = ztest_random(2); /* * Periodically change the zfs_abd_scatter_enabled setting. */ if (ztest_random(10) == 0) zfs_abd_scatter_enabled = ztest_random(2); } thread_exit(); } static __attribute__((noreturn)) void ztest_deadman_thread(void *arg) { ztest_shared_t *zs = arg; spa_t *spa = ztest_spa; hrtime_t delay, overdue, last_run = gethrtime(); delay = (zs->zs_thread_stop - zs->zs_thread_start) + MSEC2NSEC(zfs_deadman_synctime_ms); while (!ztest_exiting) { /* * Wait for the delay timer while checking occasionally * if we should stop. */ if (gethrtime() < last_run + delay) { (void) poll(NULL, 0, 1000); continue; } /* * If the pool is suspended then fail immediately. Otherwise, * check to see if the pool is making any progress. If * vdev_deadman() discovers that there hasn't been any recent * I/Os then it will end up aborting the tests. */ if (spa_suspended(spa) || spa->spa_root_vdev == NULL) { fatal(B_FALSE, "aborting test after %lu seconds because " "pool has transitioned to a suspended state.", zfs_deadman_synctime_ms / 1000); } vdev_deadman(spa->spa_root_vdev, FTAG); /* * If the process doesn't complete within a grace period of * zfs_deadman_synctime_ms over the expected finish time, * then it may be hung and is terminated. */ overdue = zs->zs_proc_stop + MSEC2NSEC(zfs_deadman_synctime_ms); if (gethrtime() > overdue) { fatal(B_FALSE, "aborting test after %llu seconds because " "the process is overdue for termination.", (gethrtime() - zs->zs_proc_start) / NANOSEC); } (void) printf("ztest has been running for %lld seconds\n", (gethrtime() - zs->zs_proc_start) / NANOSEC); last_run = gethrtime(); delay = MSEC2NSEC(zfs_deadman_checktime_ms); } thread_exit(); } static void ztest_execute(int test, ztest_info_t *zi, uint64_t id) { ztest_ds_t *zd = &ztest_ds[id % ztest_opts.zo_datasets]; ztest_shared_callstate_t *zc = ZTEST_GET_SHARED_CALLSTATE(test); hrtime_t functime = gethrtime(); int i; for (i = 0; i < zi->zi_iters; i++) zi->zi_func(zd, id); functime = gethrtime() - functime; atomic_add_64(&zc->zc_count, 1); atomic_add_64(&zc->zc_time, functime); if (ztest_opts.zo_verbose >= 4) (void) printf("%6.2f sec in %s\n", (double)functime / NANOSEC, zi->zi_funcname); } static __attribute__((noreturn)) void ztest_thread(void *arg) { int rand; uint64_t id = (uintptr_t)arg; ztest_shared_t *zs = ztest_shared; uint64_t call_next; hrtime_t now; ztest_info_t *zi; ztest_shared_callstate_t *zc; while ((now = gethrtime()) < zs->zs_thread_stop) { /* * See if it's time to force a crash. */ if (now > zs->zs_thread_kill) ztest_kill(zs); /* * If we're getting ENOSPC with some regularity, stop. */ if (zs->zs_enospc_count > 10) break; /* * Pick a random function to execute. */ rand = ztest_random(ZTEST_FUNCS); zi = &ztest_info[rand]; zc = ZTEST_GET_SHARED_CALLSTATE(rand); call_next = zc->zc_next; if (now >= call_next && atomic_cas_64(&zc->zc_next, call_next, call_next + ztest_random(2 * zi->zi_interval[0] + 1)) == call_next) { ztest_execute(rand, zi, id); } } thread_exit(); } static void ztest_dataset_name(char *dsname, const char *pool, int d) { (void) snprintf(dsname, ZFS_MAX_DATASET_NAME_LEN, "%s/ds_%d", pool, d); } static void ztest_dataset_destroy(int d) { char name[ZFS_MAX_DATASET_NAME_LEN]; int t; ztest_dataset_name(name, ztest_opts.zo_pool, d); if (ztest_opts.zo_verbose >= 3) (void) printf("Destroying %s to free up space\n", name); /* * Cleanup any non-standard clones and snapshots. In general, * ztest thread t operates on dataset (t % zopt_datasets), * so there may be more than one thing to clean up. */ for (t = d; t < ztest_opts.zo_threads; t += ztest_opts.zo_datasets) ztest_dsl_dataset_cleanup(name, t); (void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL, DS_FIND_SNAPSHOTS | DS_FIND_CHILDREN); } static void ztest_dataset_dirobj_verify(ztest_ds_t *zd) { uint64_t usedobjs, dirobjs, scratch; /* * ZTEST_DIROBJ is the object directory for the entire dataset. * Therefore, the number of objects in use should equal the * number of ZTEST_DIROBJ entries, +1 for ZTEST_DIROBJ itself. * If not, we have an object leak. * * Note that we can only check this in ztest_dataset_open(), * when the open-context and syncing-context values agree. * That's because zap_count() returns the open-context value, * while dmu_objset_space() returns the rootbp fill count. */ VERIFY0(zap_count(zd->zd_os, ZTEST_DIROBJ, &dirobjs)); dmu_objset_space(zd->zd_os, &scratch, &scratch, &usedobjs, &scratch); ASSERT3U(dirobjs + 1, ==, usedobjs); } static int ztest_dataset_open(int d) { ztest_ds_t *zd = &ztest_ds[d]; uint64_t committed_seq = ZTEST_GET_SHARED_DS(d)->zd_seq; objset_t *os; zilog_t *zilog; char name[ZFS_MAX_DATASET_NAME_LEN]; int error; ztest_dataset_name(name, ztest_opts.zo_pool, d); (void) pthread_rwlock_rdlock(&ztest_name_lock); error = ztest_dataset_create(name); if (error == ENOSPC) { (void) pthread_rwlock_unlock(&ztest_name_lock); ztest_record_enospc(FTAG); return (error); } ASSERT(error == 0 || error == EEXIST); VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE, zd, &os)); (void) pthread_rwlock_unlock(&ztest_name_lock); ztest_zd_init(zd, ZTEST_GET_SHARED_DS(d), os); zilog = zd->zd_zilog; if (zilog->zl_header->zh_claim_lr_seq != 0 && zilog->zl_header->zh_claim_lr_seq < committed_seq) fatal(B_FALSE, "missing log records: " "claimed %"PRIu64" < committed %"PRIu64"", zilog->zl_header->zh_claim_lr_seq, committed_seq); ztest_dataset_dirobj_verify(zd); zil_replay(os, zd, ztest_replay_vector); ztest_dataset_dirobj_verify(zd); if (ztest_opts.zo_verbose >= 6) (void) printf("%s replay %"PRIu64" blocks, " "%"PRIu64" records, seq %"PRIu64"\n", zd->zd_name, zilog->zl_parse_blk_count, zilog->zl_parse_lr_count, zilog->zl_replaying_seq); zilog = zil_open(os, ztest_get_data, NULL); if (zilog->zl_replaying_seq != 0 && zilog->zl_replaying_seq < committed_seq) fatal(B_FALSE, "missing log records: " "replayed %"PRIu64" < committed %"PRIu64"", zilog->zl_replaying_seq, committed_seq); return (0); } static void ztest_dataset_close(int d) { ztest_ds_t *zd = &ztest_ds[d]; zil_close(zd->zd_zilog); dmu_objset_disown(zd->zd_os, B_TRUE, zd); ztest_zd_fini(zd); } static int ztest_replay_zil_cb(const char *name, void *arg) { (void) arg; objset_t *os; ztest_ds_t *zdtmp; VERIFY0(ztest_dmu_objset_own(name, DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os)); zdtmp = umem_alloc(sizeof (ztest_ds_t), UMEM_NOFAIL); ztest_zd_init(zdtmp, NULL, os); zil_replay(os, zdtmp, ztest_replay_vector); ztest_zd_fini(zdtmp); if (dmu_objset_zil(os)->zl_parse_lr_count != 0 && ztest_opts.zo_verbose >= 6) { zilog_t *zilog = dmu_objset_zil(os); (void) printf("%s replay %"PRIu64" blocks, " "%"PRIu64" records, seq %"PRIu64"\n", name, zilog->zl_parse_blk_count, zilog->zl_parse_lr_count, zilog->zl_replaying_seq); } umem_free(zdtmp, sizeof (ztest_ds_t)); dmu_objset_disown(os, B_TRUE, FTAG); return (0); } static void ztest_freeze(void) { ztest_ds_t *zd = &ztest_ds[0]; spa_t *spa; int numloops = 0; if (ztest_opts.zo_verbose >= 3) (void) printf("testing spa_freeze()...\n"); kernel_init(SPA_MODE_READ | SPA_MODE_WRITE); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); VERIFY0(ztest_dataset_open(0)); ztest_spa = spa; /* * Force the first log block to be transactionally allocated. * We have to do this before we freeze the pool -- otherwise * the log chain won't be anchored. */ while (BP_IS_HOLE(&zd->zd_zilog->zl_header->zh_log)) { ztest_dmu_object_alloc_free(zd, 0); zil_commit(zd->zd_zilog, 0); } txg_wait_synced(spa_get_dsl(spa), 0); /* * Freeze the pool. This stops spa_sync() from doing anything, * so that the only way to record changes from now on is the ZIL. */ spa_freeze(spa); /* * Because it is hard to predict how much space a write will actually * require beforehand, we leave ourselves some fudge space to write over * capacity. */ uint64_t capacity = metaslab_class_get_space(spa_normal_class(spa)) / 2; /* * Run tests that generate log records but don't alter the pool config * or depend on DSL sync tasks (snapshots, objset create/destroy, etc). * We do a txg_wait_synced() after each iteration to force the txg * to increase well beyond the last synced value in the uberblock. * The ZIL should be OK with that. * * Run a random number of times less than zo_maxloops and ensure we do * not run out of space on the pool. */ while (ztest_random(10) != 0 && numloops++ < ztest_opts.zo_maxloops && metaslab_class_get_alloc(spa_normal_class(spa)) < capacity) { ztest_od_t od; ztest_od_init(&od, 0, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0); VERIFY0(ztest_object_init(zd, &od, sizeof (od), B_FALSE)); ztest_io(zd, od.od_object, ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT); txg_wait_synced(spa_get_dsl(spa), 0); } /* * Commit all of the changes we just generated. */ zil_commit(zd->zd_zilog, 0); txg_wait_synced(spa_get_dsl(spa), 0); /* * Close our dataset and close the pool. */ ztest_dataset_close(0); spa_close(spa, FTAG); kernel_fini(); /* * Open and close the pool and dataset to induce log replay. */ kernel_init(SPA_MODE_READ | SPA_MODE_WRITE); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); ASSERT3U(spa_freeze_txg(spa), ==, UINT64_MAX); VERIFY0(ztest_dataset_open(0)); ztest_spa = spa; txg_wait_synced(spa_get_dsl(spa), 0); ztest_dataset_close(0); ztest_reguid(NULL, 0); spa_close(spa, FTAG); kernel_fini(); } static void ztest_import_impl(void) { importargs_t args = { 0 }; nvlist_t *cfg = NULL; int nsearch = 1; char *searchdirs[nsearch]; int flags = ZFS_IMPORT_MISSING_LOG; searchdirs[0] = ztest_opts.zo_dir; args.paths = nsearch; args.path = searchdirs; args.can_be_active = B_FALSE; VERIFY0(zpool_find_config(NULL, ztest_opts.zo_pool, &cfg, &args, &libzpool_config_ops)); VERIFY0(spa_import(ztest_opts.zo_pool, cfg, NULL, flags)); fnvlist_free(cfg); } /* * Import a storage pool with the given name. */ static void ztest_import(ztest_shared_t *zs) { spa_t *spa; mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL)); kernel_init(SPA_MODE_READ | SPA_MODE_WRITE); ztest_import_impl(); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); zs->zs_metaslab_sz = 1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift; spa_close(spa, FTAG); kernel_fini(); if (!ztest_opts.zo_mmp_test) { ztest_run_zdb(ztest_opts.zo_pool); ztest_freeze(); ztest_run_zdb(ztest_opts.zo_pool); } (void) pthread_rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); mutex_destroy(&ztest_checkpoint_lock); } /* * Kick off threads to run tests on all datasets in parallel. */ static void ztest_run(ztest_shared_t *zs) { spa_t *spa; objset_t *os; kthread_t *resume_thread, *deadman_thread; kthread_t **run_threads; uint64_t object; int error; int t, d; ztest_exiting = B_FALSE; /* * Initialize parent/child shared state. */ mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL)); zs->zs_thread_start = gethrtime(); zs->zs_thread_stop = zs->zs_thread_start + ztest_opts.zo_passtime * NANOSEC; zs->zs_thread_stop = MIN(zs->zs_thread_stop, zs->zs_proc_stop); zs->zs_thread_kill = zs->zs_thread_stop; if (ztest_random(100) < ztest_opts.zo_killrate) { zs->zs_thread_kill -= ztest_random(ztest_opts.zo_passtime * NANOSEC); } mutex_init(&zcl.zcl_callbacks_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zcl.zcl_callbacks, sizeof (ztest_cb_data_t), offsetof(ztest_cb_data_t, zcd_node)); /* * Open our pool. It may need to be imported first depending on * what tests were running when the previous pass was terminated. */ kernel_init(SPA_MODE_READ | SPA_MODE_WRITE); error = spa_open(ztest_opts.zo_pool, &spa, FTAG); if (error) { VERIFY3S(error, ==, ENOENT); ztest_import_impl(); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); zs->zs_metaslab_sz = 1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift; } metaslab_preload_limit = ztest_random(20) + 1; ztest_spa = spa; VERIFY0(vdev_raidz_impl_set("cycle")); dmu_objset_stats_t dds; VERIFY0(ztest_dmu_objset_own(ztest_opts.zo_pool, DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os)); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); dmu_objset_fast_stat(os, &dds); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); zs->zs_guid = dds.dds_guid; dmu_objset_disown(os, B_TRUE, FTAG); /* * Create a thread to periodically resume suspended I/O. */ resume_thread = thread_create(NULL, 0, ztest_resume_thread, spa, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri); /* * Create a deadman thread and set to panic if we hang. */ deadman_thread = thread_create(NULL, 0, ztest_deadman_thread, zs, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri); spa->spa_deadman_failmode = ZIO_FAILURE_MODE_PANIC; /* * Verify that we can safely inquire about any object, * whether it's allocated or not. To make it interesting, * we probe a 5-wide window around each power of two. * This hits all edge cases, including zero and the max. */ for (t = 0; t < 64; t++) { for (d = -5; d <= 5; d++) { error = dmu_object_info(spa->spa_meta_objset, (1ULL << t) + d, NULL); ASSERT(error == 0 || error == ENOENT || error == EINVAL); } } /* * If we got any ENOSPC errors on the previous run, destroy something. */ if (zs->zs_enospc_count != 0) { int d = ztest_random(ztest_opts.zo_datasets); ztest_dataset_destroy(d); } zs->zs_enospc_count = 0; /* * If we were in the middle of ztest_device_removal() and were killed * we need to ensure the removal and scrub complete before running * any tests that check ztest_device_removal_active. The removal will * be restarted automatically when the spa is opened, but we need to * initiate the scrub manually if it is not already in progress. Note * that we always run the scrub whenever an indirect vdev exists * because we have no way of knowing for sure if ztest_device_removal() * fully completed its scrub before the pool was reimported. */ if (spa->spa_removing_phys.sr_state == DSS_SCANNING || spa->spa_removing_phys.sr_prev_indirect_vdev != -1) { while (spa->spa_removing_phys.sr_state == DSS_SCANNING) txg_wait_synced(spa_get_dsl(spa), 0); error = ztest_scrub_impl(spa); if (error == EBUSY) error = 0; ASSERT0(error); } run_threads = umem_zalloc(ztest_opts.zo_threads * sizeof (kthread_t *), UMEM_NOFAIL); if (ztest_opts.zo_verbose >= 4) (void) printf("starting main threads...\n"); /* * Replay all logs of all datasets in the pool. This is primarily for * temporary datasets which wouldn't otherwise get replayed, which * can trigger failures when attempting to offline a SLOG in * ztest_fault_inject(). */ (void) dmu_objset_find(ztest_opts.zo_pool, ztest_replay_zil_cb, NULL, DS_FIND_CHILDREN); /* * Kick off all the tests that run in parallel. */ for (t = 0; t < ztest_opts.zo_threads; t++) { if (t < ztest_opts.zo_datasets && ztest_dataset_open(t) != 0) { umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *)); return; } run_threads[t] = thread_create(NULL, 0, ztest_thread, (void *)(uintptr_t)t, 0, NULL, TS_RUN | TS_JOINABLE, defclsyspri); } /* * Wait for all of the tests to complete. */ for (t = 0; t < ztest_opts.zo_threads; t++) VERIFY0(thread_join(run_threads[t])); /* * Close all datasets. This must be done after all the threads * are joined so we can be sure none of the datasets are in-use * by any of the threads. */ for (t = 0; t < ztest_opts.zo_threads; t++) { if (t < ztest_opts.zo_datasets) ztest_dataset_close(t); } txg_wait_synced(spa_get_dsl(spa), 0); zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa)); zs->zs_space = metaslab_class_get_space(spa_normal_class(spa)); umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *)); /* Kill the resume and deadman threads */ ztest_exiting = B_TRUE; VERIFY0(thread_join(resume_thread)); VERIFY0(thread_join(deadman_thread)); ztest_resume(spa); /* * Right before closing the pool, kick off a bunch of async I/O; * spa_close() should wait for it to complete. */ for (object = 1; object < 50; object++) { dmu_prefetch(spa->spa_meta_objset, object, 0, 0, 1ULL << 20, ZIO_PRIORITY_SYNC_READ); } /* Verify that at least one commit cb was called in a timely fashion */ if (zc_cb_counter >= ZTEST_COMMIT_CB_MIN_REG) VERIFY0(zc_min_txg_delay); spa_close(spa, FTAG); /* * Verify that we can loop over all pools. */ mutex_enter(&spa_namespace_lock); for (spa = spa_next(NULL); spa != NULL; spa = spa_next(spa)) if (ztest_opts.zo_verbose > 3) (void) printf("spa_next: found %s\n", spa_name(spa)); mutex_exit(&spa_namespace_lock); /* * Verify that we can export the pool and reimport it under a * different name. */ if ((ztest_random(2) == 0) && !ztest_opts.zo_mmp_test) { char name[ZFS_MAX_DATASET_NAME_LEN]; (void) snprintf(name, sizeof (name), "%s_import", ztest_opts.zo_pool); ztest_spa_import_export(ztest_opts.zo_pool, name); ztest_spa_import_export(name, ztest_opts.zo_pool); } kernel_fini(); list_destroy(&zcl.zcl_callbacks); mutex_destroy(&zcl.zcl_callbacks_lock); (void) pthread_rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); mutex_destroy(&ztest_checkpoint_lock); } static void print_time(hrtime_t t, char *timebuf) { hrtime_t s = t / NANOSEC; hrtime_t m = s / 60; hrtime_t h = m / 60; hrtime_t d = h / 24; s -= m * 60; m -= h * 60; h -= d * 24; timebuf[0] = '\0'; if (d) (void) sprintf(timebuf, "%llud%02lluh%02llum%02llus", d, h, m, s); else if (h) (void) sprintf(timebuf, "%lluh%02llum%02llus", h, m, s); else if (m) (void) sprintf(timebuf, "%llum%02llus", m, s); else (void) sprintf(timebuf, "%llus", s); } static nvlist_t * make_random_props(void) { nvlist_t *props; props = fnvlist_alloc(); if (ztest_random(2) == 0) return (props); fnvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_AUTOREPLACE), 1); return (props); } /* * Create a storage pool with the given name and initial vdev size. * Then test spa_freeze() functionality. */ static void ztest_init(ztest_shared_t *zs) { spa_t *spa; nvlist_t *nvroot, *props; int i; mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL); VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL)); kernel_init(SPA_MODE_READ | SPA_MODE_WRITE); /* * Create the storage pool. */ (void) spa_destroy(ztest_opts.zo_pool); ztest_shared->zs_vdev_next_leaf = 0; zs->zs_splits = 0; zs->zs_mirrors = ztest_opts.zo_mirrors; nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0, NULL, ztest_opts.zo_raid_children, zs->zs_mirrors, 1); props = make_random_props(); /* * We don't expect the pool to suspend unless maxfaults == 0, * in which case ztest_fault_inject() temporarily takes away * the only valid replica. */ fnvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_FAILUREMODE), MAXFAULTS(zs) ? ZIO_FAILURE_MODE_PANIC : ZIO_FAILURE_MODE_WAIT); for (i = 0; i < SPA_FEATURES; i++) { char *buf; if (!spa_feature_table[i].fi_zfs_mod_supported) continue; /* * 75% chance of using the log space map feature. We want ztest * to exercise both the code paths that use the log space map * feature and the ones that don't. */ if (i == SPA_FEATURE_LOG_SPACEMAP && ztest_random(4) == 0) continue; VERIFY3S(-1, !=, asprintf(&buf, "feature@%s", spa_feature_table[i].fi_uname)); fnvlist_add_uint64(props, buf, 0); free(buf); } VERIFY0(spa_create(ztest_opts.zo_pool, nvroot, props, NULL, NULL)); fnvlist_free(nvroot); fnvlist_free(props); VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG)); zs->zs_metaslab_sz = 1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift; spa_close(spa, FTAG); kernel_fini(); if (!ztest_opts.zo_mmp_test) { ztest_run_zdb(ztest_opts.zo_pool); ztest_freeze(); ztest_run_zdb(ztest_opts.zo_pool); } (void) pthread_rwlock_destroy(&ztest_name_lock); mutex_destroy(&ztest_vdev_lock); mutex_destroy(&ztest_checkpoint_lock); } static void setup_data_fd(void) { static char ztest_name_data[] = "/tmp/ztest.data.XXXXXX"; ztest_fd_data = mkstemp(ztest_name_data); ASSERT3S(ztest_fd_data, >=, 0); (void) unlink(ztest_name_data); } static int shared_data_size(ztest_shared_hdr_t *hdr) { int size; size = hdr->zh_hdr_size; size += hdr->zh_opts_size; size += hdr->zh_size; size += hdr->zh_stats_size * hdr->zh_stats_count; size += hdr->zh_ds_size * hdr->zh_ds_count; return (size); } static void setup_hdr(void) { int size; ztest_shared_hdr_t *hdr; hdr = (void *)mmap(0, P2ROUNDUP(sizeof (*hdr), getpagesize()), PROT_READ | PROT_WRITE, MAP_SHARED, ztest_fd_data, 0); ASSERT3P(hdr, !=, MAP_FAILED); VERIFY0(ftruncate(ztest_fd_data, sizeof (ztest_shared_hdr_t))); hdr->zh_hdr_size = sizeof (ztest_shared_hdr_t); hdr->zh_opts_size = sizeof (ztest_shared_opts_t); hdr->zh_size = sizeof (ztest_shared_t); hdr->zh_stats_size = sizeof (ztest_shared_callstate_t); hdr->zh_stats_count = ZTEST_FUNCS; hdr->zh_ds_size = sizeof (ztest_shared_ds_t); hdr->zh_ds_count = ztest_opts.zo_datasets; size = shared_data_size(hdr); VERIFY0(ftruncate(ztest_fd_data, size)); (void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize())); } static void setup_data(void) { int size, offset; ztest_shared_hdr_t *hdr; uint8_t *buf; hdr = (void *)mmap(0, P2ROUNDUP(sizeof (*hdr), getpagesize()), PROT_READ, MAP_SHARED, ztest_fd_data, 0); ASSERT3P(hdr, !=, MAP_FAILED); size = shared_data_size(hdr); (void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize())); hdr = ztest_shared_hdr = (void *)mmap(0, P2ROUNDUP(size, getpagesize()), PROT_READ | PROT_WRITE, MAP_SHARED, ztest_fd_data, 0); ASSERT3P(hdr, !=, MAP_FAILED); buf = (uint8_t *)hdr; offset = hdr->zh_hdr_size; ztest_shared_opts = (void *)&buf[offset]; offset += hdr->zh_opts_size; ztest_shared = (void *)&buf[offset]; offset += hdr->zh_size; ztest_shared_callstate = (void *)&buf[offset]; offset += hdr->zh_stats_size * hdr->zh_stats_count; ztest_shared_ds = (void *)&buf[offset]; } static boolean_t exec_child(char *cmd, char *libpath, boolean_t ignorekill, int *statusp) { pid_t pid; int status; char *cmdbuf = NULL; pid = fork(); if (cmd == NULL) { cmdbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL); (void) strlcpy(cmdbuf, getexecname(), MAXPATHLEN); cmd = cmdbuf; } if (pid == -1) fatal(B_TRUE, "fork failed"); if (pid == 0) { /* child */ char fd_data_str[12]; VERIFY3S(11, >=, snprintf(fd_data_str, 12, "%d", ztest_fd_data)); VERIFY0(setenv("ZTEST_FD_DATA", fd_data_str, 1)); if (libpath != NULL) { const char *curlp = getenv("LD_LIBRARY_PATH"); if (curlp == NULL) VERIFY0(setenv("LD_LIBRARY_PATH", libpath, 1)); else { char *newlp = NULL; VERIFY3S(-1, !=, asprintf(&newlp, "%s:%s", libpath, curlp)); VERIFY0(setenv("LD_LIBRARY_PATH", newlp, 1)); free(newlp); } } (void) execl(cmd, cmd, (char *)NULL); ztest_dump_core = B_FALSE; fatal(B_TRUE, "exec failed: %s", cmd); } if (cmdbuf != NULL) { umem_free(cmdbuf, MAXPATHLEN); cmd = NULL; } while (waitpid(pid, &status, 0) != pid) continue; if (statusp != NULL) *statusp = status; if (WIFEXITED(status)) { if (WEXITSTATUS(status) != 0) { (void) fprintf(stderr, "child exited with code %d\n", WEXITSTATUS(status)); exit(2); } return (B_FALSE); } else if (WIFSIGNALED(status)) { if (!ignorekill || WTERMSIG(status) != SIGKILL) { (void) fprintf(stderr, "child died with signal %d\n", WTERMSIG(status)); exit(3); } return (B_TRUE); } else { (void) fprintf(stderr, "something strange happened to child\n"); exit(4); } } static void ztest_run_init(void) { int i; ztest_shared_t *zs = ztest_shared; /* * Blow away any existing copy of zpool.cache */ (void) remove(spa_config_path); if (ztest_opts.zo_init == 0) { if (ztest_opts.zo_verbose >= 1) (void) printf("Importing pool %s\n", ztest_opts.zo_pool); ztest_import(zs); return; } /* * Create and initialize our storage pool. */ for (i = 1; i <= ztest_opts.zo_init; i++) { memset(zs, 0, sizeof (*zs)); if (ztest_opts.zo_verbose >= 3 && ztest_opts.zo_init != 1) { (void) printf("ztest_init(), pass %d\n", i); } ztest_init(zs); } } int main(int argc, char **argv) { int kills = 0; int iters = 0; int older = 0; int newer = 0; ztest_shared_t *zs; ztest_info_t *zi; ztest_shared_callstate_t *zc; char timebuf[100]; char numbuf[NN_NUMBUF_SZ]; char *cmd; boolean_t hasalt; int f, err; char *fd_data_str = getenv("ZTEST_FD_DATA"); struct sigaction action; (void) setvbuf(stdout, NULL, _IOLBF, 0); dprintf_setup(&argc, argv); zfs_deadman_synctime_ms = 300000; zfs_deadman_checktime_ms = 30000; /* * As two-word space map entries may not come up often (especially * if pool and vdev sizes are small) we want to force at least some * of them so the feature get tested. */ zfs_force_some_double_word_sm_entries = B_TRUE; /* * Verify that even extensively damaged split blocks with many * segments can be reconstructed in a reasonable amount of time * when reconstruction is known to be possible. * * Note: the lower this value is, the more damage we inflict, and * the more time ztest spends in recovering that damage. We chose * to induce damage 1/100th of the time so recovery is tested but * not so frequently that ztest doesn't get to test other code paths. */ zfs_reconstruct_indirect_damage_fraction = 100; action.sa_handler = sig_handler; sigemptyset(&action.sa_mask); action.sa_flags = 0; if (sigaction(SIGSEGV, &action, NULL) < 0) { (void) fprintf(stderr, "ztest: cannot catch SIGSEGV: %s.\n", strerror(errno)); exit(EXIT_FAILURE); } if (sigaction(SIGABRT, &action, NULL) < 0) { (void) fprintf(stderr, "ztest: cannot catch SIGABRT: %s.\n", strerror(errno)); exit(EXIT_FAILURE); } /* * Force random_get_bytes() to use /dev/urandom in order to prevent * ztest from needlessly depleting the system entropy pool. */ random_path = "/dev/urandom"; ztest_fd_rand = open(random_path, O_RDONLY | O_CLOEXEC); ASSERT3S(ztest_fd_rand, >=, 0); if (!fd_data_str) { process_options(argc, argv); setup_data_fd(); setup_hdr(); setup_data(); memcpy(ztest_shared_opts, &ztest_opts, sizeof (*ztest_shared_opts)); } else { ztest_fd_data = atoi(fd_data_str); setup_data(); memcpy(&ztest_opts, ztest_shared_opts, sizeof (ztest_opts)); } ASSERT3U(ztest_opts.zo_datasets, ==, ztest_shared_hdr->zh_ds_count); err = ztest_set_global_vars(); if (err != 0 && !fd_data_str) { /* error message done by ztest_set_global_vars */ exit(EXIT_FAILURE); } else { /* children should not be spawned if setting gvars fails */ VERIFY3S(err, ==, 0); } /* Override location of zpool.cache */ VERIFY3S(asprintf((char **)&spa_config_path, "%s/zpool.cache", ztest_opts.zo_dir), !=, -1); ztest_ds = umem_alloc(ztest_opts.zo_datasets * sizeof (ztest_ds_t), UMEM_NOFAIL); zs = ztest_shared; if (fd_data_str) { metaslab_force_ganging = ztest_opts.zo_metaslab_force_ganging; metaslab_df_alloc_threshold = zs->zs_metaslab_df_alloc_threshold; if (zs->zs_do_init) ztest_run_init(); else ztest_run(zs); exit(0); } hasalt = (strlen(ztest_opts.zo_alt_ztest) != 0); if (ztest_opts.zo_verbose >= 1) { (void) printf("%"PRIu64" vdevs, %d datasets, %d threads," "%d %s disks, %"PRIu64" seconds...\n\n", ztest_opts.zo_vdevs, ztest_opts.zo_datasets, ztest_opts.zo_threads, ztest_opts.zo_raid_children, ztest_opts.zo_raid_type, ztest_opts.zo_time); } cmd = umem_alloc(MAXNAMELEN, UMEM_NOFAIL); (void) strlcpy(cmd, getexecname(), MAXNAMELEN); zs->zs_do_init = B_TRUE; if (strlen(ztest_opts.zo_alt_ztest) != 0) { if (ztest_opts.zo_verbose >= 1) { (void) printf("Executing older ztest for " "initialization: %s\n", ztest_opts.zo_alt_ztest); } VERIFY(!exec_child(ztest_opts.zo_alt_ztest, ztest_opts.zo_alt_libpath, B_FALSE, NULL)); } else { VERIFY(!exec_child(NULL, NULL, B_FALSE, NULL)); } zs->zs_do_init = B_FALSE; zs->zs_proc_start = gethrtime(); zs->zs_proc_stop = zs->zs_proc_start + ztest_opts.zo_time * NANOSEC; for (f = 0; f < ZTEST_FUNCS; f++) { zi = &ztest_info[f]; zc = ZTEST_GET_SHARED_CALLSTATE(f); if (zs->zs_proc_start + zi->zi_interval[0] > zs->zs_proc_stop) zc->zc_next = UINT64_MAX; else zc->zc_next = zs->zs_proc_start + ztest_random(2 * zi->zi_interval[0] + 1); } /* * Run the tests in a loop. These tests include fault injection * to verify that self-healing data works, and forced crashes * to verify that we never lose on-disk consistency. */ while (gethrtime() < zs->zs_proc_stop) { int status; boolean_t killed; /* * Initialize the workload counters for each function. */ for (f = 0; f < ZTEST_FUNCS; f++) { zc = ZTEST_GET_SHARED_CALLSTATE(f); zc->zc_count = 0; zc->zc_time = 0; } /* Set the allocation switch size */ zs->zs_metaslab_df_alloc_threshold = ztest_random(zs->zs_metaslab_sz / 4) + 1; if (!hasalt || ztest_random(2) == 0) { if (hasalt && ztest_opts.zo_verbose >= 1) { (void) printf("Executing newer ztest: %s\n", cmd); } newer++; killed = exec_child(cmd, NULL, B_TRUE, &status); } else { if (hasalt && ztest_opts.zo_verbose >= 1) { (void) printf("Executing older ztest: %s\n", ztest_opts.zo_alt_ztest); } older++; killed = exec_child(ztest_opts.zo_alt_ztest, ztest_opts.zo_alt_libpath, B_TRUE, &status); } if (killed) kills++; iters++; if (ztest_opts.zo_verbose >= 1) { hrtime_t now = gethrtime(); now = MIN(now, zs->zs_proc_stop); print_time(zs->zs_proc_stop - now, timebuf); nicenum(zs->zs_space, numbuf, sizeof (numbuf)); (void) printf("Pass %3d, %8s, %3"PRIu64" ENOSPC, " "%4.1f%% of %5s used, %3.0f%% done, %8s to go\n", iters, WIFEXITED(status) ? "Complete" : "SIGKILL", zs->zs_enospc_count, 100.0 * zs->zs_alloc / zs->zs_space, numbuf, 100.0 * (now - zs->zs_proc_start) / (ztest_opts.zo_time * NANOSEC), timebuf); } if (ztest_opts.zo_verbose >= 2) { (void) printf("\nWorkload summary:\n\n"); (void) printf("%7s %9s %s\n", "Calls", "Time", "Function"); (void) printf("%7s %9s %s\n", "-----", "----", "--------"); for (f = 0; f < ZTEST_FUNCS; f++) { zi = &ztest_info[f]; zc = ZTEST_GET_SHARED_CALLSTATE(f); print_time(zc->zc_time, timebuf); (void) printf("%7"PRIu64" %9s %s\n", zc->zc_count, timebuf, zi->zi_funcname); } (void) printf("\n"); } if (!ztest_opts.zo_mmp_test) ztest_run_zdb(ztest_opts.zo_pool); } if (ztest_opts.zo_verbose >= 1) { if (hasalt) { (void) printf("%d runs of older ztest: %s\n", older, ztest_opts.zo_alt_ztest); (void) printf("%d runs of newer ztest: %s\n", newer, cmd); } (void) printf("%d killed, %d completed, %.0f%% kill rate\n", kills, iters - kills, (100.0 * kills) / MAX(1, iters)); } umem_free(cmd, MAXNAMELEN); return (0); } diff --git a/config/kernel-blkdev.m4 b/config/kernel-blkdev.m4 index 462d6c6efa8e..28e5364581ea 100644 --- a/config/kernel-blkdev.m4 +++ b/config/kernel-blkdev.m4 @@ -1,424 +1,479 @@ dnl # dnl # 2.6.38 API change, dnl # Added blkdev_get_by_path() dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH], [ ZFS_LINUX_TEST_SRC([blkdev_get_by_path], [ #include #include ], [ struct block_device *bdev __attribute__ ((unused)) = NULL; const char *path = "path"; fmode_t mode = 0; void *holder = NULL; bdev = blkdev_get_by_path(path, mode, holder); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH], [ AC_MSG_CHECKING([whether blkdev_get_by_path() exists]) ZFS_LINUX_TEST_RESULT([blkdev_get_by_path], [ AC_MSG_RESULT(yes) ], [ ZFS_LINUX_TEST_ERROR([blkdev_get_by_path()]) ]) ]) dnl # dnl # 2.6.38 API change, dnl # Added blkdev_put() dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_PUT], [ ZFS_LINUX_TEST_SRC([blkdev_put], [ #include #include ], [ struct block_device *bdev = NULL; fmode_t mode = 0; blkdev_put(bdev, mode); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_PUT], [ AC_MSG_CHECKING([whether blkdev_put() exists]) ZFS_LINUX_TEST_RESULT([blkdev_put], [ AC_MSG_RESULT(yes) ], [ ZFS_LINUX_TEST_ERROR([blkdev_put()]) ]) ]) dnl # dnl # 4.1 API, exported blkdev_reread_part() symbol, back ported to the dnl # 3.10.0 CentOS 7.x enterprise kernels. dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_REREAD_PART], [ ZFS_LINUX_TEST_SRC([blkdev_reread_part], [ #include #include ], [ struct block_device *bdev = NULL; int error; error = blkdev_reread_part(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_REREAD_PART], [ AC_MSG_CHECKING([whether blkdev_reread_part() exists]) ZFS_LINUX_TEST_RESULT([blkdev_reread_part], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BLKDEV_REREAD_PART, 1, [blkdev_reread_part() exists]) ], [ AC_MSG_RESULT(no) ]) ]) dnl # dnl # check_disk_change() was removed in 5.10 dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_CHECK_DISK_CHANGE], [ ZFS_LINUX_TEST_SRC([check_disk_change], [ #include #include ], [ struct block_device *bdev = NULL; bool error; error = check_disk_change(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_CHECK_DISK_CHANGE], [ AC_MSG_CHECKING([whether check_disk_change() exists]) ZFS_LINUX_TEST_RESULT([check_disk_change], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_CHECK_DISK_CHANGE, 1, [check_disk_change() exists]) ], [ AC_MSG_RESULT(no) ]) ]) +dnl # +dnl # bdev_kobj() is introduced from 5.12 +dnl # +AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_KOBJ], [ + ZFS_LINUX_TEST_SRC([bdev_kobj], [ + #include + #include + #include + ], [ + struct block_device *bdev = NULL; + struct kobject *disk_kobj; + disk_kobj = bdev_kobj(bdev); + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_KOBJ], [ + AC_MSG_CHECKING([whether bdev_kobj() exists]) + ZFS_LINUX_TEST_RESULT([bdev_kobj], [ + AC_MSG_RESULT(yes) + AC_DEFINE(HAVE_BDEV_KOBJ, 1, + [bdev_kobj() exists]) + ], [ + AC_MSG_RESULT(no) + ]) +]) + +dnl # +dnl # part_to_dev() was removed in 5.12 +dnl # +AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_PART_TO_DEV], [ + ZFS_LINUX_TEST_SRC([part_to_dev], [ + #include + #include + ], [ + struct hd_struct *p = NULL; + struct device *pdev; + pdev = part_to_dev(p); + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_PART_TO_DEV], [ + AC_MSG_CHECKING([whether part_to_dev() exists]) + ZFS_LINUX_TEST_RESULT([part_to_dev], [ + AC_MSG_RESULT(yes) + AC_DEFINE(HAVE_PART_TO_DEV, 1, + [part_to_dev() exists]) + ], [ + AC_MSG_RESULT(no) + ]) +]) + dnl # dnl # 5.10 API, check_disk_change() is removed, in favor of dnl # bdev_check_media_change(), which doesn't force revalidation dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE], [ ZFS_LINUX_TEST_SRC([bdev_check_media_change], [ #include #include ], [ struct block_device *bdev = NULL; int error; error = bdev_check_media_change(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_CHECK_MEDIA_CHANGE], [ AC_MSG_CHECKING([whether bdev_check_media_change() exists]) ZFS_LINUX_TEST_RESULT([bdev_check_media_change], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BDEV_CHECK_MEDIA_CHANGE, 1, [bdev_check_media_change() exists]) ], [ AC_MSG_RESULT(no) ]) ]) dnl # dnl # 2.6.22 API change dnl # Single argument invalidate_bdev() dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_INVALIDATE_BDEV], [ ZFS_LINUX_TEST_SRC([invalidate_bdev], [ #include #include ],[ struct block_device *bdev = NULL; invalidate_bdev(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_INVALIDATE_BDEV], [ AC_MSG_CHECKING([whether invalidate_bdev() exists]) ZFS_LINUX_TEST_RESULT([invalidate_bdev], [ AC_MSG_RESULT(yes) ],[ ZFS_LINUX_TEST_ERROR([invalidate_bdev()]) ]) ]) dnl # dnl # 5.11 API, lookup_bdev() takes dev_t argument. dnl # 2.6.27 API, lookup_bdev() was first exported. dnl # 4.4.0-6.21 API, lookup_bdev() on Ubuntu takes mode argument. dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_LOOKUP_BDEV], [ ZFS_LINUX_TEST_SRC([lookup_bdev_devt], [ #include ], [ int error __attribute__ ((unused)); const char path[] = "/example/path"; dev_t dev; error = lookup_bdev(path, &dev); ]) ZFS_LINUX_TEST_SRC([lookup_bdev_1arg], [ #include #include ], [ struct block_device *bdev __attribute__ ((unused)); const char path[] = "/example/path"; bdev = lookup_bdev(path); ]) ZFS_LINUX_TEST_SRC([lookup_bdev_mode], [ #include ], [ struct block_device *bdev __attribute__ ((unused)); const char path[] = "/example/path"; bdev = lookup_bdev(path, FMODE_READ); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_LOOKUP_BDEV], [ AC_MSG_CHECKING([whether lookup_bdev() wants dev_t arg]) ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_devt], [lookup_bdev], [fs/block_dev.c], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_DEVT_LOOKUP_BDEV, 1, [lookup_bdev() wants dev_t arg]) ], [ AC_MSG_RESULT(no) AC_MSG_CHECKING([whether lookup_bdev() wants 1 arg]) ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_1arg], [lookup_bdev], [fs/block_dev.c], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_1ARG_LOOKUP_BDEV, 1, [lookup_bdev() wants 1 arg]) ], [ AC_MSG_RESULT(no) AC_MSG_CHECKING([whether lookup_bdev() wants mode arg]) ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_mode], [lookup_bdev], [fs/block_dev.c], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_MODE_LOOKUP_BDEV, 1, [lookup_bdev() wants mode arg]) ], [ ZFS_LINUX_TEST_ERROR([lookup_bdev()]) ]) ]) ]) ]) dnl # dnl # 2.6.30 API change dnl # dnl # The bdev_physical_block_size() interface was added to provide a way dnl # to determine the smallest write which can be performed without a dnl # read-modify-write operation. dnl # dnl # Unfortunately, this interface isn't entirely reliable because dnl # drives are sometimes known to misreport this value. dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE], [ ZFS_LINUX_TEST_SRC([bdev_physical_block_size], [ #include ],[ struct block_device *bdev __attribute__ ((unused)) = NULL; bdev_physical_block_size(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE], [ AC_MSG_CHECKING([whether bdev_physical_block_size() is available]) ZFS_LINUX_TEST_RESULT([bdev_physical_block_size], [ AC_MSG_RESULT(yes) ],[ ZFS_LINUX_TEST_ERROR([bdev_physical_block_size()]) ]) ]) dnl # dnl # 2.6.30 API change dnl # Added bdev_logical_block_size(). dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE], [ ZFS_LINUX_TEST_SRC([bdev_logical_block_size], [ #include ],[ struct block_device *bdev __attribute__ ((unused)) = NULL; bdev_logical_block_size(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE], [ AC_MSG_CHECKING([whether bdev_logical_block_size() is available]) ZFS_LINUX_TEST_RESULT([bdev_logical_block_size], [ AC_MSG_RESULT(yes) ],[ ZFS_LINUX_TEST_ERROR([bdev_logical_block_size()]) ]) ]) dnl # dnl # 5.11 API change dnl # Added bdev_whole() helper. dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE], [ ZFS_LINUX_TEST_SRC([bdev_whole], [ #include ],[ struct block_device *bdev = NULL; bdev = bdev_whole(bdev); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE], [ AC_MSG_CHECKING([whether bdev_whole() is available]) ZFS_LINUX_TEST_RESULT([bdev_whole], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BDEV_WHOLE, 1, [bdev_whole() is available]) ],[ AC_MSG_RESULT(no) ]) ]) dnl # dnl # 5.20 API change, dnl # Removed bdevname(), snprintf(.., %pg) should be used. dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEVNAME], [ ZFS_LINUX_TEST_SRC([bdevname], [ #include #include ], [ struct block_device *bdev __attribute__ ((unused)) = NULL; char path[BDEVNAME_SIZE]; (void) bdevname(bdev, path); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEVNAME], [ AC_MSG_CHECKING([whether bdevname() exists]) ZFS_LINUX_TEST_RESULT([bdevname], [ AC_DEFINE(HAVE_BDEVNAME, 1, [bdevname() is available]) AC_MSG_RESULT(yes) ], [ AC_MSG_RESULT(no) ]) ]) dnl # dnl # 5.19 API: blkdev_issue_secure_erase() dnl # 3.10 API: blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE) dnl # AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE], [ ZFS_LINUX_TEST_SRC([blkdev_issue_secure_erase], [ #include ],[ struct block_device *bdev = NULL; sector_t sector = 0; sector_t nr_sects = 0; int error __attribute__ ((unused)); error = blkdev_issue_secure_erase(bdev, sector, nr_sects, GFP_KERNEL); ]) ZFS_LINUX_TEST_SRC([blkdev_issue_discard_flags], [ #include ],[ struct block_device *bdev = NULL; sector_t sector = 0; sector_t nr_sects = 0; unsigned long flags = 0; int error __attribute__ ((unused)); error = blkdev_issue_discard(bdev, sector, nr_sects, GFP_KERNEL, flags); ]) ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE], [ AC_MSG_CHECKING([whether blkdev_issue_secure_erase() is available]) ZFS_LINUX_TEST_RESULT([blkdev_issue_secure_erase], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BLKDEV_ISSUE_SECURE_ERASE, 1, [blkdev_issue_secure_erase() is available]) ],[ AC_MSG_RESULT(no) AC_MSG_CHECKING([whether blkdev_issue_discard() is available]) ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_flags], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD, 1, [blkdev_issue_discard() is available]) ],[ ZFS_LINUX_TEST_ERROR([blkdev_issue_discard()]) ]) ]) ]) dnl # dnl # 5.13 API change dnl # blkdev_get_by_path() no longer handles ERESTARTSYS dnl # dnl # Unfortunately we're forced to rely solely on the kernel version dnl # number in order to determine the expected behavior. This was an dnl # internal change to blkdev_get_by_dev(), see commit a8ed1a0607. dnl # AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS], [ AC_MSG_CHECKING([whether blkdev_get_by_path() handles ERESTARTSYS]) AS_VERSION_COMPARE([$LINUX_VERSION], [5.13.0], [ AC_MSG_RESULT(yes) AC_DEFINE(HAVE_BLKDEV_GET_ERESTARTSYS, 1, [blkdev_get_by_path() handles ERESTARTSYS]) ],[ AC_MSG_RESULT(no) ],[ AC_MSG_RESULT(no) ]) ]) AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV], [ ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH ZFS_AC_KERNEL_SRC_BLKDEV_PUT ZFS_AC_KERNEL_SRC_BLKDEV_REREAD_PART ZFS_AC_KERNEL_SRC_BLKDEV_INVALIDATE_BDEV ZFS_AC_KERNEL_SRC_BLKDEV_LOOKUP_BDEV ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE ZFS_AC_KERNEL_SRC_BLKDEV_CHECK_DISK_CHANGE ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE ZFS_AC_KERNEL_SRC_BLKDEV_BDEVNAME ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE + ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_KOBJ + ZFS_AC_KERNEL_SRC_BLKDEV_PART_TO_DEV ]) AC_DEFUN([ZFS_AC_KERNEL_BLKDEV], [ ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH ZFS_AC_KERNEL_BLKDEV_PUT ZFS_AC_KERNEL_BLKDEV_REREAD_PART ZFS_AC_KERNEL_BLKDEV_INVALIDATE_BDEV ZFS_AC_KERNEL_BLKDEV_LOOKUP_BDEV ZFS_AC_KERNEL_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE ZFS_AC_KERNEL_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE ZFS_AC_KERNEL_BLKDEV_CHECK_DISK_CHANGE ZFS_AC_KERNEL_BLKDEV_BDEV_CHECK_MEDIA_CHANGE ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE ZFS_AC_KERNEL_BLKDEV_BDEVNAME ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE + ZFS_AC_KERNEL_BLKDEV_BDEV_KOBJ + ZFS_AC_KERNEL_BLKDEV_PART_TO_DEV ]) diff --git a/include/libzfs.h b/include/libzfs.h index 9bd4613c1091..df17873369ad 100644 --- a/include/libzfs.h +++ b/include/libzfs.h @@ -1,1002 +1,1003 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2022 by Delphix. All rights reserved. * Copyright Joyent, Inc. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2016, Intel Corporation. * Copyright 2016 Nexenta Systems, Inc. * Copyright (c) 2017 Open-E, Inc. All Rights Reserved. * Copyright (c) 2019 Datto Inc. * Copyright (c) 2021, Colm Buckley */ #ifndef _LIBZFS_H #define _LIBZFS_H extern __attribute__((visibility("default"))) #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Miscellaneous ZFS constants */ #define ZFS_MAXPROPLEN MAXPATHLEN #define ZPOOL_MAXPROPLEN MAXPATHLEN /* * libzfs errors */ typedef enum zfs_error { EZFS_SUCCESS = 0, /* no error -- success */ EZFS_NOMEM = 2000, /* out of memory */ EZFS_BADPROP, /* invalid property value */ EZFS_PROPREADONLY, /* cannot set readonly property */ EZFS_PROPTYPE, /* property does not apply to dataset type */ EZFS_PROPNONINHERIT, /* property is not inheritable */ EZFS_PROPSPACE, /* bad quota or reservation */ EZFS_BADTYPE, /* dataset is not of appropriate type */ EZFS_BUSY, /* pool or dataset is busy */ EZFS_EXISTS, /* pool or dataset already exists */ EZFS_NOENT, /* no such pool or dataset */ EZFS_BADSTREAM, /* bad backup stream */ EZFS_DSREADONLY, /* dataset is readonly */ EZFS_VOLTOOBIG, /* volume is too large for 32-bit system */ EZFS_INVALIDNAME, /* invalid dataset name */ EZFS_BADRESTORE, /* unable to restore to destination */ EZFS_BADBACKUP, /* backup failed */ EZFS_BADTARGET, /* bad attach/detach/replace target */ EZFS_NODEVICE, /* no such device in pool */ EZFS_BADDEV, /* invalid device to add */ EZFS_NOREPLICAS, /* no valid replicas */ EZFS_RESILVERING, /* resilvering (healing reconstruction) */ EZFS_BADVERSION, /* unsupported version */ EZFS_POOLUNAVAIL, /* pool is currently unavailable */ EZFS_DEVOVERFLOW, /* too many devices in one vdev */ EZFS_BADPATH, /* must be an absolute path */ EZFS_CROSSTARGET, /* rename or clone across pool or dataset */ EZFS_ZONED, /* used improperly in local zone */ EZFS_MOUNTFAILED, /* failed to mount dataset */ EZFS_UMOUNTFAILED, /* failed to unmount dataset */ EZFS_UNSHARENFSFAILED, /* failed to unshare over nfs */ EZFS_SHARENFSFAILED, /* failed to share over nfs */ EZFS_PERM, /* permission denied */ EZFS_NOSPC, /* out of space */ EZFS_FAULT, /* bad address */ EZFS_IO, /* I/O error */ EZFS_INTR, /* signal received */ EZFS_ISSPARE, /* device is a hot spare */ EZFS_INVALCONFIG, /* invalid vdev configuration */ EZFS_RECURSIVE, /* recursive dependency */ EZFS_NOHISTORY, /* no history object */ EZFS_POOLPROPS, /* couldn't retrieve pool props */ EZFS_POOL_NOTSUP, /* ops not supported for this type of pool */ EZFS_POOL_INVALARG, /* invalid argument for this pool operation */ EZFS_NAMETOOLONG, /* dataset name is too long */ EZFS_OPENFAILED, /* open of device failed */ EZFS_NOCAP, /* couldn't get capacity */ EZFS_LABELFAILED, /* write of label failed */ EZFS_BADWHO, /* invalid permission who */ EZFS_BADPERM, /* invalid permission */ EZFS_BADPERMSET, /* invalid permission set name */ EZFS_NODELEGATION, /* delegated administration is disabled */ EZFS_UNSHARESMBFAILED, /* failed to unshare over smb */ EZFS_SHARESMBFAILED, /* failed to share over smb */ EZFS_BADCACHE, /* bad cache file */ EZFS_ISL2CACHE, /* device is for the level 2 ARC */ EZFS_VDEVNOTSUP, /* unsupported vdev type */ EZFS_NOTSUP, /* ops not supported on this dataset */ EZFS_ACTIVE_SPARE, /* pool has active shared spare devices */ EZFS_UNPLAYED_LOGS, /* log device has unplayed logs */ EZFS_REFTAG_RELE, /* snapshot release: tag not found */ EZFS_REFTAG_HOLD, /* snapshot hold: tag already exists */ EZFS_TAGTOOLONG, /* snapshot hold/rele: tag too long */ EZFS_PIPEFAILED, /* pipe create failed */ EZFS_THREADCREATEFAILED, /* thread create failed */ EZFS_POSTSPLIT_ONLINE, /* onlining a disk after splitting it */ EZFS_SCRUBBING, /* currently scrubbing */ EZFS_NO_SCRUB, /* no active scrub */ EZFS_DIFF, /* general failure of zfs diff */ EZFS_DIFFDATA, /* bad zfs diff data */ EZFS_POOLREADONLY, /* pool is in read-only mode */ EZFS_SCRUB_PAUSED, /* scrub currently paused */ EZFS_ACTIVE_POOL, /* pool is imported on a different system */ EZFS_CRYPTOFAILED, /* failed to setup encryption */ EZFS_NO_PENDING, /* cannot cancel, no operation is pending */ EZFS_CHECKPOINT_EXISTS, /* checkpoint exists */ EZFS_DISCARDING_CHECKPOINT, /* currently discarding a checkpoint */ EZFS_NO_CHECKPOINT, /* pool has no checkpoint */ EZFS_DEVRM_IN_PROGRESS, /* a device is currently being removed */ EZFS_VDEV_TOO_BIG, /* a device is too big to be used */ EZFS_IOC_NOTSUPPORTED, /* operation not supported by zfs module */ EZFS_TOOMANY, /* argument list too long */ EZFS_INITIALIZING, /* currently initializing */ EZFS_NO_INITIALIZE, /* no active initialize */ EZFS_WRONG_PARENT, /* invalid parent dataset (e.g ZVOL) */ EZFS_TRIMMING, /* currently trimming */ EZFS_NO_TRIM, /* no active trim */ EZFS_TRIM_NOTSUP, /* device does not support trim */ EZFS_NO_RESILVER_DEFER, /* pool doesn't support resilver_defer */ EZFS_EXPORT_IN_PROGRESS, /* currently exporting the pool */ EZFS_REBUILDING, /* resilvering (sequential reconstrution) */ EZFS_VDEV_NOTSUP, /* ops not supported for this type of vdev */ EZFS_NOT_USER_NAMESPACE, /* a file is not a user namespace */ EZFS_CKSUM, /* insufficient replicas */ EZFS_RESUME_EXISTS, /* Resume on existing dataset without force */ EZFS_UNKNOWN } zfs_error_t; /* * The following data structures are all part * of the zfs_allow_t data structure which is * used for printing 'allow' permissions. * It is a linked list of zfs_allow_t's which * then contain avl tree's for user/group/sets/... * and each one of the entries in those trees have * avl tree's for the permissions they belong to and * whether they are local,descendent or local+descendent * permissions. The AVL trees are used primarily for * sorting purposes, but also so that we can quickly find * a given user and or permission. */ typedef struct zfs_perm_node { avl_node_t z_node; char z_pname[MAXPATHLEN]; } zfs_perm_node_t; typedef struct zfs_allow_node { avl_node_t z_node; char z_key[MAXPATHLEN]; /* name, such as joe */ avl_tree_t z_localdescend; /* local+descendent perms */ avl_tree_t z_local; /* local permissions */ avl_tree_t z_descend; /* descendent permissions */ } zfs_allow_node_t; typedef struct zfs_allow { struct zfs_allow *z_next; char z_setpoint[MAXPATHLEN]; avl_tree_t z_sets; avl_tree_t z_crperms; avl_tree_t z_user; avl_tree_t z_group; avl_tree_t z_everyone; } zfs_allow_t; /* * Basic handle types */ typedef struct zfs_handle zfs_handle_t; typedef struct zpool_handle zpool_handle_t; typedef struct libzfs_handle libzfs_handle_t; _LIBZFS_H int zpool_wait(zpool_handle_t *, zpool_wait_activity_t); _LIBZFS_H int zpool_wait_status(zpool_handle_t *, zpool_wait_activity_t, boolean_t *, boolean_t *); /* * Library initialization */ _LIBZFS_H libzfs_handle_t *libzfs_init(void); _LIBZFS_H void libzfs_fini(libzfs_handle_t *); _LIBZFS_H libzfs_handle_t *zpool_get_handle(zpool_handle_t *); _LIBZFS_H libzfs_handle_t *zfs_get_handle(zfs_handle_t *); _LIBZFS_H void libzfs_print_on_error(libzfs_handle_t *, boolean_t); _LIBZFS_H void zfs_save_arguments(int argc, char **, char *, int); _LIBZFS_H int zpool_log_history(libzfs_handle_t *, const char *); _LIBZFS_H int libzfs_errno(libzfs_handle_t *); _LIBZFS_H const char *libzfs_error_init(int); _LIBZFS_H const char *libzfs_error_action(libzfs_handle_t *); _LIBZFS_H const char *libzfs_error_description(libzfs_handle_t *); _LIBZFS_H int zfs_standard_error(libzfs_handle_t *, int, const char *); _LIBZFS_H void libzfs_mnttab_init(libzfs_handle_t *); _LIBZFS_H void libzfs_mnttab_fini(libzfs_handle_t *); _LIBZFS_H void libzfs_mnttab_cache(libzfs_handle_t *, boolean_t); _LIBZFS_H int libzfs_mnttab_find(libzfs_handle_t *, const char *, struct mnttab *); _LIBZFS_H void libzfs_mnttab_add(libzfs_handle_t *, const char *, const char *, const char *); _LIBZFS_H void libzfs_mnttab_remove(libzfs_handle_t *, const char *); /* * Basic handle functions */ _LIBZFS_H zpool_handle_t *zpool_open(libzfs_handle_t *, const char *); _LIBZFS_H zpool_handle_t *zpool_open_canfail(libzfs_handle_t *, const char *); _LIBZFS_H void zpool_close(zpool_handle_t *); _LIBZFS_H const char *zpool_get_name(zpool_handle_t *); _LIBZFS_H int zpool_get_state(zpool_handle_t *); _LIBZFS_H const char *zpool_state_to_name(vdev_state_t, vdev_aux_t); _LIBZFS_H const char *zpool_pool_state_to_name(pool_state_t); _LIBZFS_H void zpool_free_handles(libzfs_handle_t *); /* * Iterate over all active pools in the system. */ typedef int (*zpool_iter_f)(zpool_handle_t *, void *); _LIBZFS_H int zpool_iter(libzfs_handle_t *, zpool_iter_f, void *); _LIBZFS_H boolean_t zpool_skip_pool(const char *); /* * Functions to create and destroy pools */ _LIBZFS_H int zpool_create(libzfs_handle_t *, const char *, nvlist_t *, nvlist_t *, nvlist_t *); _LIBZFS_H int zpool_destroy(zpool_handle_t *, const char *); _LIBZFS_H int zpool_add(zpool_handle_t *, nvlist_t *); typedef struct splitflags { /* do not split, but return the config that would be split off */ int dryrun : 1; /* after splitting, import the pool */ int import : 1; int name_flags; } splitflags_t; typedef struct trimflags { /* requested vdevs are for the entire pool */ boolean_t fullpool; /* request a secure trim, requires support from device */ boolean_t secure; /* after starting trim, block until trim completes */ boolean_t wait; /* trim at the requested rate in bytes/second */ uint64_t rate; } trimflags_t; /* * Functions to manipulate pool and vdev state */ _LIBZFS_H int zpool_scan(zpool_handle_t *, pool_scan_func_t, pool_scrub_cmd_t); _LIBZFS_H int zpool_initialize(zpool_handle_t *, pool_initialize_func_t, nvlist_t *); _LIBZFS_H int zpool_initialize_wait(zpool_handle_t *, pool_initialize_func_t, nvlist_t *); _LIBZFS_H int zpool_trim(zpool_handle_t *, pool_trim_func_t, nvlist_t *, trimflags_t *); _LIBZFS_H int zpool_clear(zpool_handle_t *, const char *, nvlist_t *); _LIBZFS_H int zpool_reguid(zpool_handle_t *); _LIBZFS_H int zpool_reopen_one(zpool_handle_t *, void *); _LIBZFS_H int zpool_sync_one(zpool_handle_t *, void *); _LIBZFS_H int zpool_vdev_online(zpool_handle_t *, const char *, int, vdev_state_t *); _LIBZFS_H int zpool_vdev_offline(zpool_handle_t *, const char *, boolean_t); _LIBZFS_H int zpool_vdev_attach(zpool_handle_t *, const char *, const char *, nvlist_t *, int, boolean_t); _LIBZFS_H int zpool_vdev_detach(zpool_handle_t *, const char *); _LIBZFS_H int zpool_vdev_remove(zpool_handle_t *, const char *); _LIBZFS_H int zpool_vdev_remove_cancel(zpool_handle_t *); _LIBZFS_H int zpool_vdev_indirect_size(zpool_handle_t *, const char *, uint64_t *); _LIBZFS_H int zpool_vdev_split(zpool_handle_t *, char *, nvlist_t **, nvlist_t *, splitflags_t); +_LIBZFS_H int zpool_vdev_remove_wanted(zpool_handle_t *, const char *); _LIBZFS_H int zpool_vdev_fault(zpool_handle_t *, uint64_t, vdev_aux_t); _LIBZFS_H int zpool_vdev_degrade(zpool_handle_t *, uint64_t, vdev_aux_t); _LIBZFS_H int zpool_vdev_clear(zpool_handle_t *, uint64_t); _LIBZFS_H nvlist_t *zpool_find_vdev(zpool_handle_t *, const char *, boolean_t *, boolean_t *, boolean_t *); _LIBZFS_H nvlist_t *zpool_find_vdev_by_physpath(zpool_handle_t *, const char *, boolean_t *, boolean_t *, boolean_t *); _LIBZFS_H int zpool_label_disk(libzfs_handle_t *, zpool_handle_t *, const char *); _LIBZFS_H uint64_t zpool_vdev_path_to_guid(zpool_handle_t *zhp, const char *path); _LIBZFS_H const char *zpool_get_state_str(zpool_handle_t *); /* * Functions to manage pool properties */ _LIBZFS_H int zpool_set_prop(zpool_handle_t *, const char *, const char *); _LIBZFS_H int zpool_get_prop(zpool_handle_t *, zpool_prop_t, char *, size_t proplen, zprop_source_t *, boolean_t literal); _LIBZFS_H uint64_t zpool_get_prop_int(zpool_handle_t *, zpool_prop_t, zprop_source_t *); _LIBZFS_H int zpool_props_refresh(zpool_handle_t *); _LIBZFS_H const char *zpool_prop_to_name(zpool_prop_t); _LIBZFS_H const char *zpool_prop_values(zpool_prop_t); /* * Functions to manage vdev properties */ _LIBZFS_H int zpool_get_vdev_prop_value(nvlist_t *, vdev_prop_t, char *, char *, size_t, zprop_source_t *, boolean_t); _LIBZFS_H int zpool_get_vdev_prop(zpool_handle_t *, const char *, vdev_prop_t, char *, char *, size_t, zprop_source_t *, boolean_t); _LIBZFS_H int zpool_get_all_vdev_props(zpool_handle_t *, const char *, nvlist_t **); _LIBZFS_H int zpool_set_vdev_prop(zpool_handle_t *, const char *, const char *, const char *); _LIBZFS_H const char *vdev_prop_to_name(vdev_prop_t); _LIBZFS_H const char *vdev_prop_values(vdev_prop_t); _LIBZFS_H boolean_t vdev_prop_user(const char *name); _LIBZFS_H const char *vdev_prop_column_name(vdev_prop_t); _LIBZFS_H boolean_t vdev_prop_align_right(vdev_prop_t); /* * Pool health statistics. */ typedef enum { /* * The following correspond to faults as defined in the (fault.fs.zfs.*) * event namespace. Each is associated with a corresponding message ID. * This must be kept in sync with the zfs_msgid_table in * lib/libzfs/libzfs_status.c. */ ZPOOL_STATUS_CORRUPT_CACHE, /* corrupt /kernel/drv/zpool.cache */ ZPOOL_STATUS_MISSING_DEV_R, /* missing device with replicas */ ZPOOL_STATUS_MISSING_DEV_NR, /* missing device with no replicas */ ZPOOL_STATUS_CORRUPT_LABEL_R, /* bad device label with replicas */ ZPOOL_STATUS_CORRUPT_LABEL_NR, /* bad device label with no replicas */ ZPOOL_STATUS_BAD_GUID_SUM, /* sum of device guids didn't match */ ZPOOL_STATUS_CORRUPT_POOL, /* pool metadata is corrupted */ ZPOOL_STATUS_CORRUPT_DATA, /* data errors in user (meta)data */ ZPOOL_STATUS_FAILING_DEV, /* device experiencing errors */ ZPOOL_STATUS_VERSION_NEWER, /* newer on-disk version */ ZPOOL_STATUS_HOSTID_MISMATCH, /* last accessed by another system */ ZPOOL_STATUS_HOSTID_ACTIVE, /* currently active on another system */ ZPOOL_STATUS_HOSTID_REQUIRED, /* multihost=on and hostid=0 */ ZPOOL_STATUS_IO_FAILURE_WAIT, /* failed I/O, failmode 'wait' */ ZPOOL_STATUS_IO_FAILURE_CONTINUE, /* failed I/O, failmode 'continue' */ ZPOOL_STATUS_IO_FAILURE_MMP, /* failed MMP, failmode not 'panic' */ ZPOOL_STATUS_BAD_LOG, /* cannot read log chain(s) */ ZPOOL_STATUS_ERRATA, /* informational errata available */ /* * If the pool has unsupported features but can still be opened in * read-only mode, its status is ZPOOL_STATUS_UNSUP_FEAT_WRITE. If the * pool has unsupported features but cannot be opened at all, its * status is ZPOOL_STATUS_UNSUP_FEAT_READ. */ ZPOOL_STATUS_UNSUP_FEAT_READ, /* unsupported features for read */ ZPOOL_STATUS_UNSUP_FEAT_WRITE, /* unsupported features for write */ /* * These faults have no corresponding message ID. At the time we are * checking the status, the original reason for the FMA fault (I/O or * checksum errors) has been lost. */ ZPOOL_STATUS_FAULTED_DEV_R, /* faulted device with replicas */ ZPOOL_STATUS_FAULTED_DEV_NR, /* faulted device with no replicas */ /* * The following are not faults per se, but still an error possibly * requiring administrative attention. There is no corresponding * message ID. */ ZPOOL_STATUS_VERSION_OLDER, /* older legacy on-disk version */ ZPOOL_STATUS_FEAT_DISABLED, /* supported features are disabled */ ZPOOL_STATUS_RESILVERING, /* device being resilvered */ ZPOOL_STATUS_OFFLINE_DEV, /* device offline */ ZPOOL_STATUS_REMOVED_DEV, /* removed device */ ZPOOL_STATUS_REBUILDING, /* device being rebuilt */ ZPOOL_STATUS_REBUILD_SCRUB, /* recommend scrubbing the pool */ ZPOOL_STATUS_NON_NATIVE_ASHIFT, /* (e.g. 512e dev with ashift of 9) */ ZPOOL_STATUS_COMPATIBILITY_ERR, /* bad 'compatibility' property */ ZPOOL_STATUS_INCOMPATIBLE_FEAT, /* feature set outside compatibility */ /* * Finally, the following indicates a healthy pool. */ ZPOOL_STATUS_OK } zpool_status_t; _LIBZFS_H zpool_status_t zpool_get_status(zpool_handle_t *, const char **, zpool_errata_t *); _LIBZFS_H zpool_status_t zpool_import_status(nvlist_t *, const char **, zpool_errata_t *); /* * Statistics and configuration functions. */ _LIBZFS_H nvlist_t *zpool_get_config(zpool_handle_t *, nvlist_t **); _LIBZFS_H nvlist_t *zpool_get_features(zpool_handle_t *); _LIBZFS_H int zpool_refresh_stats(zpool_handle_t *, boolean_t *); _LIBZFS_H int zpool_get_errlog(zpool_handle_t *, nvlist_t **); /* * Import and export functions */ _LIBZFS_H int zpool_export(zpool_handle_t *, boolean_t, const char *); _LIBZFS_H int zpool_export_force(zpool_handle_t *, const char *); _LIBZFS_H int zpool_import(libzfs_handle_t *, nvlist_t *, const char *, char *altroot); _LIBZFS_H int zpool_import_props(libzfs_handle_t *, nvlist_t *, const char *, nvlist_t *, int); _LIBZFS_H void zpool_print_unsup_feat(nvlist_t *config); /* * Miscellaneous pool functions */ struct zfs_cmd; _LIBZFS_H const char *const zfs_history_event_names[]; typedef enum { VDEV_NAME_PATH = 1 << 0, VDEV_NAME_GUID = 1 << 1, VDEV_NAME_FOLLOW_LINKS = 1 << 2, VDEV_NAME_TYPE_ID = 1 << 3, } vdev_name_t; _LIBZFS_H char *zpool_vdev_name(libzfs_handle_t *, zpool_handle_t *, nvlist_t *, int name_flags); _LIBZFS_H int zpool_upgrade(zpool_handle_t *, uint64_t); _LIBZFS_H int zpool_get_history(zpool_handle_t *, nvlist_t **, uint64_t *, boolean_t *); _LIBZFS_H int zpool_events_next(libzfs_handle_t *, nvlist_t **, int *, unsigned, int); _LIBZFS_H int zpool_events_clear(libzfs_handle_t *, int *); _LIBZFS_H int zpool_events_seek(libzfs_handle_t *, uint64_t, int); _LIBZFS_H void zpool_obj_to_path_ds(zpool_handle_t *, uint64_t, uint64_t, char *, size_t); _LIBZFS_H void zpool_obj_to_path(zpool_handle_t *, uint64_t, uint64_t, char *, size_t); _LIBZFS_H int zfs_ioctl(libzfs_handle_t *, int, struct zfs_cmd *); _LIBZFS_H void zpool_explain_recover(libzfs_handle_t *, const char *, int, nvlist_t *); _LIBZFS_H int zpool_checkpoint(zpool_handle_t *); _LIBZFS_H int zpool_discard_checkpoint(zpool_handle_t *); _LIBZFS_H boolean_t zpool_is_draid_spare(const char *); /* * Basic handle manipulations. These functions do not create or destroy the * underlying datasets, only the references to them. */ _LIBZFS_H zfs_handle_t *zfs_open(libzfs_handle_t *, const char *, int); _LIBZFS_H zfs_handle_t *zfs_handle_dup(zfs_handle_t *); _LIBZFS_H void zfs_close(zfs_handle_t *); _LIBZFS_H zfs_type_t zfs_get_type(const zfs_handle_t *); _LIBZFS_H zfs_type_t zfs_get_underlying_type(const zfs_handle_t *); _LIBZFS_H const char *zfs_get_name(const zfs_handle_t *); _LIBZFS_H zpool_handle_t *zfs_get_pool_handle(const zfs_handle_t *); _LIBZFS_H const char *zfs_get_pool_name(const zfs_handle_t *); /* * Property management functions. Some functions are shared with the kernel, * and are found in sys/fs/zfs.h. */ /* * zfs dataset property management */ _LIBZFS_H const char *zfs_prop_default_string(zfs_prop_t); _LIBZFS_H uint64_t zfs_prop_default_numeric(zfs_prop_t); _LIBZFS_H const char *zfs_prop_column_name(zfs_prop_t); _LIBZFS_H boolean_t zfs_prop_align_right(zfs_prop_t); _LIBZFS_H nvlist_t *zfs_valid_proplist(libzfs_handle_t *, zfs_type_t, nvlist_t *, uint64_t, zfs_handle_t *, zpool_handle_t *, boolean_t, const char *); _LIBZFS_H const char *zfs_prop_to_name(zfs_prop_t); _LIBZFS_H int zfs_prop_set(zfs_handle_t *, const char *, const char *); _LIBZFS_H int zfs_prop_set_list(zfs_handle_t *, nvlist_t *); _LIBZFS_H int zfs_prop_get(zfs_handle_t *, zfs_prop_t, char *, size_t, zprop_source_t *, char *, size_t, boolean_t); _LIBZFS_H int zfs_prop_get_recvd(zfs_handle_t *, const char *, char *, size_t, boolean_t); _LIBZFS_H int zfs_prop_get_numeric(zfs_handle_t *, zfs_prop_t, uint64_t *, zprop_source_t *, char *, size_t); _LIBZFS_H int zfs_prop_get_userquota_int(zfs_handle_t *zhp, const char *propname, uint64_t *propvalue); _LIBZFS_H int zfs_prop_get_userquota(zfs_handle_t *zhp, const char *propname, char *propbuf, int proplen, boolean_t literal); _LIBZFS_H int zfs_prop_get_written_int(zfs_handle_t *zhp, const char *propname, uint64_t *propvalue); _LIBZFS_H int zfs_prop_get_written(zfs_handle_t *zhp, const char *propname, char *propbuf, int proplen, boolean_t literal); _LIBZFS_H int zfs_prop_get_feature(zfs_handle_t *zhp, const char *propname, char *buf, size_t len); _LIBZFS_H uint64_t getprop_uint64(zfs_handle_t *, zfs_prop_t, char **); _LIBZFS_H uint64_t zfs_prop_get_int(zfs_handle_t *, zfs_prop_t); _LIBZFS_H int zfs_prop_inherit(zfs_handle_t *, const char *, boolean_t); _LIBZFS_H const char *zfs_prop_values(zfs_prop_t); _LIBZFS_H int zfs_prop_is_string(zfs_prop_t prop); _LIBZFS_H nvlist_t *zfs_get_all_props(zfs_handle_t *); _LIBZFS_H nvlist_t *zfs_get_user_props(zfs_handle_t *); _LIBZFS_H nvlist_t *zfs_get_recvd_props(zfs_handle_t *); _LIBZFS_H nvlist_t *zfs_get_clones_nvl(zfs_handle_t *); _LIBZFS_H int zfs_wait_status(zfs_handle_t *, zfs_wait_activity_t, boolean_t *, boolean_t *); /* * zfs encryption management */ _LIBZFS_H int zfs_crypto_get_encryption_root(zfs_handle_t *, boolean_t *, char *); _LIBZFS_H int zfs_crypto_create(libzfs_handle_t *, char *, nvlist_t *, nvlist_t *, boolean_t stdin_available, uint8_t **, uint_t *); _LIBZFS_H int zfs_crypto_clone_check(libzfs_handle_t *, zfs_handle_t *, char *, nvlist_t *); _LIBZFS_H int zfs_crypto_attempt_load_keys(libzfs_handle_t *, const char *); _LIBZFS_H int zfs_crypto_load_key(zfs_handle_t *, boolean_t, const char *); _LIBZFS_H int zfs_crypto_unload_key(zfs_handle_t *); _LIBZFS_H int zfs_crypto_rewrap(zfs_handle_t *, nvlist_t *, boolean_t); typedef struct zprop_list { int pl_prop; char *pl_user_prop; struct zprop_list *pl_next; boolean_t pl_all; size_t pl_width; size_t pl_recvd_width; boolean_t pl_fixed; } zprop_list_t; _LIBZFS_H int zfs_expand_proplist(zfs_handle_t *, zprop_list_t **, boolean_t, boolean_t); _LIBZFS_H void zfs_prune_proplist(zfs_handle_t *, uint8_t *); _LIBZFS_H int vdev_expand_proplist(zpool_handle_t *, const char *, zprop_list_t **); #define ZFS_MOUNTPOINT_NONE "none" #define ZFS_MOUNTPOINT_LEGACY "legacy" #define ZFS_FEATURE_DISABLED "disabled" #define ZFS_FEATURE_ENABLED "enabled" #define ZFS_FEATURE_ACTIVE "active" #define ZFS_UNSUPPORTED_INACTIVE "inactive" #define ZFS_UNSUPPORTED_READONLY "readonly" /* * zpool property management */ _LIBZFS_H int zpool_expand_proplist(zpool_handle_t *, zprop_list_t **, zfs_type_t, boolean_t); _LIBZFS_H int zpool_prop_get_feature(zpool_handle_t *, const char *, char *, size_t); _LIBZFS_H const char *zpool_prop_default_string(zpool_prop_t); _LIBZFS_H uint64_t zpool_prop_default_numeric(zpool_prop_t); _LIBZFS_H const char *zpool_prop_column_name(zpool_prop_t); _LIBZFS_H boolean_t zpool_prop_align_right(zpool_prop_t); /* * Functions shared by zfs and zpool property management. */ _LIBZFS_H int zprop_iter(zprop_func func, void *cb, boolean_t show_all, boolean_t ordered, zfs_type_t type); _LIBZFS_H int zprop_get_list(libzfs_handle_t *, char *, zprop_list_t **, zfs_type_t); _LIBZFS_H void zprop_free_list(zprop_list_t *); #define ZFS_GET_NCOLS 5 typedef enum { GET_COL_NONE, GET_COL_NAME, GET_COL_PROPERTY, GET_COL_VALUE, GET_COL_RECVD, GET_COL_SOURCE } zfs_get_column_t; /* * Functions for printing zfs or zpool properties */ typedef struct vdev_cbdata { int cb_name_flags; char **cb_names; unsigned int cb_names_count; } vdev_cbdata_t; typedef struct zprop_get_cbdata { int cb_sources; zfs_get_column_t cb_columns[ZFS_GET_NCOLS]; int cb_colwidths[ZFS_GET_NCOLS + 1]; boolean_t cb_scripted; boolean_t cb_literal; boolean_t cb_first; zprop_list_t *cb_proplist; zfs_type_t cb_type; vdev_cbdata_t cb_vdevs; } zprop_get_cbdata_t; _LIBZFS_H void zprop_print_one_property(const char *, zprop_get_cbdata_t *, const char *, const char *, zprop_source_t, const char *, const char *); /* * Iterator functions. */ typedef int (*zfs_iter_f)(zfs_handle_t *, void *); _LIBZFS_H int zfs_iter_root(libzfs_handle_t *, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_children(zfs_handle_t *, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_dependents(zfs_handle_t *, boolean_t, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_filesystems(zfs_handle_t *, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_snapshots(zfs_handle_t *, boolean_t, zfs_iter_f, void *, uint64_t, uint64_t); _LIBZFS_H int zfs_iter_snapshots_sorted(zfs_handle_t *, zfs_iter_f, void *, uint64_t, uint64_t); _LIBZFS_H int zfs_iter_snapspec(zfs_handle_t *, const char *, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_bookmarks(zfs_handle_t *, zfs_iter_f, void *); _LIBZFS_H int zfs_iter_mounted(zfs_handle_t *, zfs_iter_f, void *); typedef struct get_all_cb { zfs_handle_t **cb_handles; size_t cb_alloc; size_t cb_used; } get_all_cb_t; _LIBZFS_H void zfs_foreach_mountpoint(libzfs_handle_t *, zfs_handle_t **, size_t, zfs_iter_f, void *, boolean_t); _LIBZFS_H void libzfs_add_handle(get_all_cb_t *, zfs_handle_t *); /* * Functions to create and destroy datasets. */ _LIBZFS_H int zfs_create(libzfs_handle_t *, const char *, zfs_type_t, nvlist_t *); _LIBZFS_H int zfs_create_ancestors(libzfs_handle_t *, const char *); _LIBZFS_H int zfs_destroy(zfs_handle_t *, boolean_t); _LIBZFS_H int zfs_destroy_snaps(zfs_handle_t *, char *, boolean_t); _LIBZFS_H int zfs_destroy_snaps_nvl(libzfs_handle_t *, nvlist_t *, boolean_t); _LIBZFS_H int zfs_destroy_snaps_nvl_os(libzfs_handle_t *, nvlist_t *); _LIBZFS_H int zfs_clone(zfs_handle_t *, const char *, nvlist_t *); _LIBZFS_H int zfs_snapshot(libzfs_handle_t *, const char *, boolean_t, nvlist_t *); _LIBZFS_H int zfs_snapshot_nvl(libzfs_handle_t *hdl, nvlist_t *snaps, nvlist_t *props); _LIBZFS_H int zfs_rollback(zfs_handle_t *, zfs_handle_t *, boolean_t); typedef struct renameflags { /* recursive rename */ int recursive : 1; /* don't unmount file systems */ int nounmount : 1; /* force unmount file systems */ int forceunmount : 1; } renameflags_t; _LIBZFS_H int zfs_rename(zfs_handle_t *, const char *, renameflags_t); typedef struct sendflags { /* Amount of extra information to print. */ int verbosity; /* recursive send (ie, -R) */ boolean_t replicate; /* for recursive send, skip sending missing snapshots */ boolean_t skipmissing; /* for incrementals, do all intermediate snapshots */ boolean_t doall; /* if dataset is a clone, do incremental from its origin */ boolean_t fromorigin; /* field no longer used, maintained for backwards compatibility */ boolean_t pad; /* send properties (ie, -p) */ boolean_t props; /* do not send (no-op, ie. -n) */ boolean_t dryrun; /* parsable verbose output (ie. -P) */ boolean_t parsable; /* show progress (ie. -v) */ boolean_t progress; /* large blocks (>128K) are permitted */ boolean_t largeblock; /* WRITE_EMBEDDED records of type DATA are permitted */ boolean_t embed_data; /* compressed WRITE records are permitted */ boolean_t compress; /* raw encrypted records are permitted */ boolean_t raw; /* only send received properties (ie. -b) */ boolean_t backup; /* include snapshot holds in send stream */ boolean_t holds; /* stream represents a partially received dataset */ boolean_t saved; } sendflags_t; typedef boolean_t (snapfilter_cb_t)(zfs_handle_t *, void *); _LIBZFS_H int zfs_send(zfs_handle_t *, const char *, const char *, sendflags_t *, int, snapfilter_cb_t, void *, nvlist_t **); _LIBZFS_H int zfs_send_one(zfs_handle_t *, const char *, int, sendflags_t *, const char *); _LIBZFS_H int zfs_send_progress(zfs_handle_t *, int, uint64_t *, uint64_t *); _LIBZFS_H int zfs_send_resume(libzfs_handle_t *, sendflags_t *, int outfd, const char *); _LIBZFS_H int zfs_send_saved(zfs_handle_t *, sendflags_t *, int, const char *); _LIBZFS_H nvlist_t *zfs_send_resume_token_to_nvlist(libzfs_handle_t *hdl, const char *token); _LIBZFS_H int zfs_promote(zfs_handle_t *); _LIBZFS_H int zfs_hold(zfs_handle_t *, const char *, const char *, boolean_t, int); _LIBZFS_H int zfs_hold_nvl(zfs_handle_t *, int, nvlist_t *); _LIBZFS_H int zfs_release(zfs_handle_t *, const char *, const char *, boolean_t); _LIBZFS_H int zfs_get_holds(zfs_handle_t *, nvlist_t **); _LIBZFS_H uint64_t zvol_volsize_to_reservation(zpool_handle_t *, uint64_t, nvlist_t *); typedef int (*zfs_userspace_cb_t)(void *arg, const char *domain, uid_t rid, uint64_t space); _LIBZFS_H int zfs_userspace(zfs_handle_t *, zfs_userquota_prop_t, zfs_userspace_cb_t, void *); _LIBZFS_H int zfs_get_fsacl(zfs_handle_t *, nvlist_t **); _LIBZFS_H int zfs_set_fsacl(zfs_handle_t *, boolean_t, nvlist_t *); typedef struct recvflags { /* print informational messages (ie, -v was specified) */ boolean_t verbose; /* the destination is a prefix, not the exact fs (ie, -d) */ boolean_t isprefix; /* * Only the tail of the sent snapshot path is appended to the * destination to determine the received snapshot name (ie, -e). */ boolean_t istail; /* do not actually do the recv, just check if it would work (ie, -n) */ boolean_t dryrun; /* rollback/destroy filesystems as necessary (eg, -F) */ boolean_t force; /* set "canmount=off" on all modified filesystems */ boolean_t canmountoff; /* * Mark the file systems as "resumable" and do not destroy them if the * receive is interrupted */ boolean_t resumable; /* byteswap flag is used internally; callers need not specify */ boolean_t byteswap; /* do not mount file systems as they are extracted (private) */ boolean_t nomount; /* Was holds flag set in the compound header? */ boolean_t holds; /* skip receive of snapshot holds */ boolean_t skipholds; /* mount the filesystem unless nomount is specified */ boolean_t domount; /* force unmount while recv snapshot (private) */ boolean_t forceunmount; /* use this recv to check (and heal if needed) an existing snapshot */ boolean_t heal; } recvflags_t; _LIBZFS_H int zfs_receive(libzfs_handle_t *, const char *, nvlist_t *, recvflags_t *, int, avl_tree_t *); typedef enum diff_flags { ZFS_DIFF_PARSEABLE = 1 << 0, ZFS_DIFF_TIMESTAMP = 1 << 1, ZFS_DIFF_CLASSIFY = 1 << 2, ZFS_DIFF_NO_MANGLE = 1 << 3 } diff_flags_t; _LIBZFS_H int zfs_show_diffs(zfs_handle_t *, int, const char *, const char *, int); /* * Miscellaneous functions. */ _LIBZFS_H const char *zfs_type_to_name(zfs_type_t); _LIBZFS_H void zfs_refresh_properties(zfs_handle_t *); _LIBZFS_H int zfs_name_valid(const char *, zfs_type_t); _LIBZFS_H zfs_handle_t *zfs_path_to_zhandle(libzfs_handle_t *, const char *, zfs_type_t); _LIBZFS_H int zfs_parent_name(zfs_handle_t *, char *, size_t); _LIBZFS_H boolean_t zfs_dataset_exists(libzfs_handle_t *, const char *, zfs_type_t); _LIBZFS_H int zfs_spa_version(zfs_handle_t *, int *); _LIBZFS_H boolean_t zfs_bookmark_exists(const char *path); /* * Mount support functions. */ _LIBZFS_H boolean_t is_mounted(libzfs_handle_t *, const char *special, char **); _LIBZFS_H boolean_t zfs_is_mounted(zfs_handle_t *, char **); _LIBZFS_H int zfs_mount(zfs_handle_t *, const char *, int); _LIBZFS_H int zfs_mount_at(zfs_handle_t *, const char *, int, const char *); _LIBZFS_H int zfs_unmount(zfs_handle_t *, const char *, int); _LIBZFS_H int zfs_unmountall(zfs_handle_t *, int); _LIBZFS_H int zfs_mount_delegation_check(void); #if defined(__linux__) || defined(__APPLE__) _LIBZFS_H int zfs_parse_mount_options(const char *mntopts, unsigned long *mntflags, unsigned long *zfsflags, int sloppy, char *badopt, char *mtabopt); _LIBZFS_H void zfs_adjust_mount_options(zfs_handle_t *zhp, const char *mntpoint, char *mntopts, char *mtabopt); #endif /* * Share support functions. * * enum sa_protocol * lists are terminated with SA_NO_PROTOCOL, * NULL means "all/any known to this libzfs". */ #define SA_NO_PROTOCOL -1 _LIBZFS_H boolean_t zfs_is_shared(zfs_handle_t *zhp, char **where, const enum sa_protocol *proto); _LIBZFS_H int zfs_share(zfs_handle_t *zhp, const enum sa_protocol *proto); _LIBZFS_H int zfs_unshare(zfs_handle_t *zhp, const char *mountpoint, const enum sa_protocol *proto); _LIBZFS_H int zfs_unshareall(zfs_handle_t *zhp, const enum sa_protocol *proto); _LIBZFS_H void zfs_commit_shares(const enum sa_protocol *proto); _LIBZFS_H void zfs_truncate_shares(const enum sa_protocol *proto); _LIBZFS_H int zfs_nicestrtonum(libzfs_handle_t *, const char *, uint64_t *); /* * Utility functions to run an external process. */ #define STDOUT_VERBOSE 0x01 #define STDERR_VERBOSE 0x02 #define NO_DEFAULT_PATH 0x04 /* Don't use $PATH to lookup the command */ _LIBZFS_H int libzfs_run_process(const char *, char **, int); _LIBZFS_H int libzfs_run_process_get_stdout(const char *, char *[], char *[], char **[], int *); _LIBZFS_H int libzfs_run_process_get_stdout_nopath(const char *, char *[], char *[], char **[], int *); _LIBZFS_H void libzfs_free_str_array(char **, int); _LIBZFS_H boolean_t libzfs_envvar_is_set(const char *); /* * Utility functions for zfs version */ _LIBZFS_H const char *zfs_version_userland(void); _LIBZFS_H char *zfs_version_kernel(void); _LIBZFS_H int zfs_version_print(void); /* * Given a device or file, determine if it is part of a pool. */ _LIBZFS_H int zpool_in_use(libzfs_handle_t *, int, pool_state_t *, char **, boolean_t *); /* * Label manipulation. */ _LIBZFS_H int zpool_clear_label(int); _LIBZFS_H int zpool_set_bootenv(zpool_handle_t *, const nvlist_t *); _LIBZFS_H int zpool_get_bootenv(zpool_handle_t *, nvlist_t **); /* * Management interfaces for SMB ACL files */ _LIBZFS_H int zfs_smb_acl_add(libzfs_handle_t *, char *, char *, char *); _LIBZFS_H int zfs_smb_acl_remove(libzfs_handle_t *, char *, char *, char *); _LIBZFS_H int zfs_smb_acl_purge(libzfs_handle_t *, char *, char *); _LIBZFS_H int zfs_smb_acl_rename(libzfs_handle_t *, char *, char *, char *, char *); /* * Enable and disable datasets within a pool by mounting/unmounting and * sharing/unsharing them. */ _LIBZFS_H int zpool_enable_datasets(zpool_handle_t *, const char *, int); _LIBZFS_H int zpool_disable_datasets(zpool_handle_t *, boolean_t); _LIBZFS_H void zpool_disable_datasets_os(zpool_handle_t *, boolean_t); _LIBZFS_H void zpool_disable_volume_os(const char *); /* * Parse a features file for -o compatibility */ typedef enum { ZPOOL_COMPATIBILITY_OK, ZPOOL_COMPATIBILITY_WARNTOKEN, ZPOOL_COMPATIBILITY_BADTOKEN, ZPOOL_COMPATIBILITY_BADFILE, ZPOOL_COMPATIBILITY_NOFILES } zpool_compat_status_t; _LIBZFS_H zpool_compat_status_t zpool_load_compat(const char *, boolean_t *, char *, size_t); #ifdef __FreeBSD__ /* * Attach/detach the given filesystem to/from the given jail. */ _LIBZFS_H int zfs_jail(zfs_handle_t *zhp, int jailid, int attach); /* * Set loader options for next boot. */ _LIBZFS_H int zpool_nextboot(libzfs_handle_t *, uint64_t, uint64_t, const char *); #endif /* __FreeBSD__ */ #ifdef __linux__ /* * Add or delete the given filesystem to/from the given user namespace. */ _LIBZFS_H int zfs_userns(zfs_handle_t *zhp, const char *nspath, int attach); #endif #ifdef __cplusplus } #endif #endif /* _LIBZFS_H */ diff --git a/include/os/linux/Makefile.am b/include/os/linux/Makefile.am index 4d6901c694c8..13ba8060c62d 100644 --- a/include/os/linux/Makefile.am +++ b/include/os/linux/Makefile.am @@ -1,111 +1,112 @@ if CONFIG_KERNEL kernel_linuxdir = $(kerneldir)/linux kernel_linux_HEADERS = \ %D%/kernel/linux/blkdev_compat.h \ %D%/kernel/linux/compiler_compat.h \ %D%/kernel/linux/dcache_compat.h \ %D%/kernel/linux/kmap_compat.h \ %D%/kernel/linux/mod_compat.h \ %D%/kernel/linux/page_compat.h \ %D%/kernel/linux/percpu_compat.h \ %D%/kernel/linux/simd.h \ %D%/kernel/linux/simd_aarch64.h \ %D%/kernel/linux/simd_powerpc.h \ %D%/kernel/linux/simd_x86.h \ %D%/kernel/linux/utsname_compat.h \ %D%/kernel/linux/vfs_compat.h \ %D%/kernel/linux/xattr_compat.h kernel_sysdir = $(kerneldir)/sys kernel_sys_HEADERS = \ %D%/zfs/sys/policy.h \ %D%/zfs/sys/sha2.h \ %D%/zfs/sys/trace_acl.h \ %D%/zfs/sys/trace_arc.h \ %D%/zfs/sys/trace_common.h \ %D%/zfs/sys/trace_dbgmsg.h \ %D%/zfs/sys/trace_dbuf.h \ %D%/zfs/sys/trace_dmu.h \ %D%/zfs/sys/trace_dnode.h \ %D%/zfs/sys/trace_multilist.h \ %D%/zfs/sys/trace_rrwlock.h \ %D%/zfs/sys/trace_txg.h \ %D%/zfs/sys/trace_vdev.h \ %D%/zfs/sys/trace_zfs.h \ %D%/zfs/sys/trace_zil.h \ %D%/zfs/sys/trace_zio.h \ %D%/zfs/sys/trace_zrlock.h \ %D%/zfs/sys/zfs_bootenv_os.h \ %D%/zfs/sys/zfs_context_os.h \ %D%/zfs/sys/zfs_ctldir.h \ %D%/zfs/sys/zfs_dir.h \ %D%/zfs/sys/zfs_vfsops_os.h \ %D%/zfs/sys/zfs_vnops_os.h \ %D%/zfs/sys/zfs_znode_impl.h \ %D%/zfs/sys/zpl.h kernel_spl_rpcdir = $(kerneldir)/spl/rpc kernel_spl_rpc_HEADERS = \ %D%/spl/rpc/xdr.h kernel_spl_sysdir = $(kerneldir)/spl/sys kernel_spl_sys_HEADERS = \ %D%/spl/sys/acl.h \ %D%/spl/sys/atomic.h \ %D%/spl/sys/byteorder.h \ %D%/spl/sys/callb.h \ %D%/spl/sys/callo.h \ %D%/spl/sys/cmn_err.h \ %D%/spl/sys/condvar.h \ %D%/spl/sys/cred.h \ %D%/spl/sys/ctype.h \ %D%/spl/sys/debug.h \ %D%/spl/sys/disp.h \ %D%/spl/sys/dkio.h \ %D%/spl/sys/errno.h \ %D%/spl/sys/fcntl.h \ %D%/spl/sys/file.h \ %D%/spl/sys/inttypes.h \ %D%/spl/sys/isa_defs.h \ %D%/spl/sys/kmem.h \ %D%/spl/sys/kmem_cache.h \ %D%/spl/sys/kstat.h \ %D%/spl/sys/list.h \ + %D%/spl/sys/misc.h \ %D%/spl/sys/mod_os.h \ %D%/spl/sys/mutex.h \ %D%/spl/sys/param.h \ %D%/spl/sys/proc.h \ %D%/spl/sys/processor.h \ %D%/spl/sys/procfs_list.h \ %D%/spl/sys/random.h \ %D%/spl/sys/rwlock.h \ %D%/spl/sys/shrinker.h \ %D%/spl/sys/sid.h \ %D%/spl/sys/signal.h \ %D%/spl/sys/simd.h \ %D%/spl/sys/stat.h \ %D%/spl/sys/string.h \ %D%/spl/sys/sunddi.h \ %D%/spl/sys/sysmacros.h \ %D%/spl/sys/systeminfo.h \ %D%/spl/sys/taskq.h \ %D%/spl/sys/thread.h \ %D%/spl/sys/time.h \ %D%/spl/sys/timer.h \ %D%/spl/sys/trace.h \ %D%/spl/sys/trace_spl.h \ %D%/spl/sys/trace_taskq.h \ %D%/spl/sys/tsd.h \ %D%/spl/sys/types.h \ %D%/spl/sys/types32.h \ %D%/spl/sys/uio.h \ %D%/spl/sys/user.h \ %D%/spl/sys/vfs.h \ %D%/spl/sys/vmem.h \ %D%/spl/sys/vmsystm.h \ %D%/spl/sys/vnode.h \ %D%/spl/sys/wait.h \ %D%/spl/sys/wmsum.h \ %D%/spl/sys/zmod.h \ %D%/spl/sys/zone.h endif diff --git a/include/os/linux/kernel/linux/blkdev_compat.h b/include/os/linux/kernel/linux/blkdev_compat.h index cdcea166903d..3276796537a4 100644 --- a/include/os/linux/kernel/linux/blkdev_compat.h +++ b/include/os/linux/kernel/linux/blkdev_compat.h @@ -1,722 +1,748 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (C) 2011 Lawrence Livermore National Security, LLC. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). * Written by Brian Behlendorf . * LLNL-CODE-403049. */ #ifndef _ZFS_BLKDEV_H #define _ZFS_BLKDEV_H #include #include #include #include #include /* for SECTOR_* */ #include #ifdef HAVE_BLK_MQ #include #endif #ifndef HAVE_BLK_QUEUE_FLAG_SET static inline void blk_queue_flag_set(unsigned int flag, struct request_queue *q) { queue_flag_set(flag, q); } #endif #ifndef HAVE_BLK_QUEUE_FLAG_CLEAR static inline void blk_queue_flag_clear(unsigned int flag, struct request_queue *q) { queue_flag_clear(flag, q); } #endif /* * 4.7 API, * The blk_queue_write_cache() interface has replaced blk_queue_flush() * interface. However, the new interface is GPL-only thus we implement * our own trivial wrapper when the GPL-only version is detected. * * 2.6.36 - 4.6 API, * The blk_queue_flush() interface has replaced blk_queue_ordered() * interface. However, while the old interface was available to all the * new one is GPL-only. Thus if the GPL-only version is detected we * implement our own trivial helper. */ static inline void blk_queue_set_write_cache(struct request_queue *q, bool wc, bool fua) { #if defined(HAVE_BLK_QUEUE_WRITE_CACHE_GPL_ONLY) if (wc) blk_queue_flag_set(QUEUE_FLAG_WC, q); else blk_queue_flag_clear(QUEUE_FLAG_WC, q); if (fua) blk_queue_flag_set(QUEUE_FLAG_FUA, q); else blk_queue_flag_clear(QUEUE_FLAG_FUA, q); #elif defined(HAVE_BLK_QUEUE_WRITE_CACHE) blk_queue_write_cache(q, wc, fua); #elif defined(HAVE_BLK_QUEUE_FLUSH_GPL_ONLY) if (wc) q->flush_flags |= REQ_FLUSH; if (fua) q->flush_flags |= REQ_FUA; #elif defined(HAVE_BLK_QUEUE_FLUSH) blk_queue_flush(q, (wc ? REQ_FLUSH : 0) | (fua ? REQ_FUA : 0)); #else #error "Unsupported kernel" #endif } static inline void blk_queue_set_read_ahead(struct request_queue *q, unsigned long ra_pages) { #if !defined(HAVE_BLK_QUEUE_UPDATE_READAHEAD) && \ !defined(HAVE_DISK_UPDATE_READAHEAD) #ifdef HAVE_BLK_QUEUE_BDI_DYNAMIC q->backing_dev_info->ra_pages = ra_pages; #else q->backing_dev_info.ra_pages = ra_pages; #endif #endif } #ifdef HAVE_BIO_BVEC_ITER #define BIO_BI_SECTOR(bio) (bio)->bi_iter.bi_sector #define BIO_BI_SIZE(bio) (bio)->bi_iter.bi_size #define BIO_BI_IDX(bio) (bio)->bi_iter.bi_idx #define BIO_BI_SKIP(bio) (bio)->bi_iter.bi_bvec_done #define bio_for_each_segment4(bv, bvp, b, i) \ bio_for_each_segment((bv), (b), (i)) typedef struct bvec_iter bvec_iterator_t; #else #define BIO_BI_SECTOR(bio) (bio)->bi_sector #define BIO_BI_SIZE(bio) (bio)->bi_size #define BIO_BI_IDX(bio) (bio)->bi_idx #define BIO_BI_SKIP(bio) (0) #define bio_for_each_segment4(bv, bvp, b, i) \ bio_for_each_segment((bvp), (b), (i)) typedef int bvec_iterator_t; #endif static inline void bio_set_flags_failfast(struct block_device *bdev, int *flags) { #ifdef CONFIG_BUG /* * Disable FAILFAST for loopback devices because of the * following incorrect BUG_ON() in loop_make_request(). * This support is also disabled for md devices because the * test suite layers md devices on top of loopback devices. * This may be removed when the loopback driver is fixed. * * BUG_ON(!lo || (rw != READ && rw != WRITE)); */ if ((MAJOR(bdev->bd_dev) == LOOP_MAJOR) || (MAJOR(bdev->bd_dev) == MD_MAJOR)) return; #ifdef BLOCK_EXT_MAJOR if (MAJOR(bdev->bd_dev) == BLOCK_EXT_MAJOR) return; #endif /* BLOCK_EXT_MAJOR */ #endif /* CONFIG_BUG */ *flags |= REQ_FAILFAST_MASK; } /* * Maximum disk label length, it may be undefined for some kernels. */ #if !defined(DISK_NAME_LEN) #define DISK_NAME_LEN 32 #endif /* DISK_NAME_LEN */ #ifdef HAVE_BIO_BI_STATUS static inline int bi_status_to_errno(blk_status_t status) { switch (status) { case BLK_STS_OK: return (0); case BLK_STS_NOTSUPP: return (EOPNOTSUPP); case BLK_STS_TIMEOUT: return (ETIMEDOUT); case BLK_STS_NOSPC: return (ENOSPC); case BLK_STS_TRANSPORT: return (ENOLINK); case BLK_STS_TARGET: return (EREMOTEIO); case BLK_STS_NEXUS: return (EBADE); case BLK_STS_MEDIUM: return (ENODATA); case BLK_STS_PROTECTION: return (EILSEQ); case BLK_STS_RESOURCE: return (ENOMEM); case BLK_STS_AGAIN: return (EAGAIN); case BLK_STS_IOERR: return (EIO); default: return (EIO); } } static inline blk_status_t errno_to_bi_status(int error) { switch (error) { case 0: return (BLK_STS_OK); case EOPNOTSUPP: return (BLK_STS_NOTSUPP); case ETIMEDOUT: return (BLK_STS_TIMEOUT); case ENOSPC: return (BLK_STS_NOSPC); case ENOLINK: return (BLK_STS_TRANSPORT); case EREMOTEIO: return (BLK_STS_TARGET); case EBADE: return (BLK_STS_NEXUS); case ENODATA: return (BLK_STS_MEDIUM); case EILSEQ: return (BLK_STS_PROTECTION); case ENOMEM: return (BLK_STS_RESOURCE); case EAGAIN: return (BLK_STS_AGAIN); case EIO: return (BLK_STS_IOERR); default: return (BLK_STS_IOERR); } } #endif /* HAVE_BIO_BI_STATUS */ /* * 4.3 API change * The bio_endio() prototype changed slightly. These are helper * macro's to ensure the prototype and invocation are handled. */ #ifdef HAVE_1ARG_BIO_END_IO_T #ifdef HAVE_BIO_BI_STATUS #define BIO_END_IO_ERROR(bio) bi_status_to_errno(bio->bi_status) #define BIO_END_IO_PROTO(fn, x, z) static void fn(struct bio *x) #define BIO_END_IO(bio, error) bio_set_bi_status(bio, error) static inline void bio_set_bi_status(struct bio *bio, int error) { ASSERT3S(error, <=, 0); bio->bi_status = errno_to_bi_status(-error); bio_endio(bio); } #else #define BIO_END_IO_ERROR(bio) (-(bio->bi_error)) #define BIO_END_IO_PROTO(fn, x, z) static void fn(struct bio *x) #define BIO_END_IO(bio, error) bio_set_bi_error(bio, error) static inline void bio_set_bi_error(struct bio *bio, int error) { ASSERT3S(error, <=, 0); bio->bi_error = error; bio_endio(bio); } #endif /* HAVE_BIO_BI_STATUS */ #else #define BIO_END_IO_PROTO(fn, x, z) static void fn(struct bio *x, int z) #define BIO_END_IO(bio, error) bio_endio(bio, error); #endif /* HAVE_1ARG_BIO_END_IO_T */ +/* + * 5.15 MACRO, + * GD_DEAD + * + * 2.6.36 - 5.14 MACRO, + * GENHD_FL_UP + * + * Check the disk status and return B_TRUE if alive + * otherwise B_FALSE + */ +static inline boolean_t +zfs_check_disk_status(struct block_device *bdev) +{ +#if defined(GENHD_FL_UP) + return (!!(bdev->bd_disk->flags & GENHD_FL_UP)); +#elif defined(GD_DEAD) + return (!test_bit(GD_DEAD, &bdev->bd_disk->state)); +#else +/* + * This is encountered if neither GENHD_FL_UP nor GD_DEAD is available in + * the kernel - likely due to an MACRO change that needs to be chased down. + */ +#error "Unsupported kernel: no usable disk status check" +#endif +} + /* * 4.1 API, * 3.10.0 CentOS 7.x API, * blkdev_reread_part() * * For older kernels trigger a re-reading of the partition table by calling * check_disk_change() which calls flush_disk() to invalidate the device. * * For newer kernels (as of 5.10), bdev_check_media_change is used, in favor of * check_disk_change(), with the modification that invalidation is no longer * forced. */ #ifdef HAVE_CHECK_DISK_CHANGE #define zfs_check_media_change(bdev) check_disk_change(bdev) #ifdef HAVE_BLKDEV_REREAD_PART #define vdev_bdev_reread_part(bdev) blkdev_reread_part(bdev) #else #define vdev_bdev_reread_part(bdev) check_disk_change(bdev) #endif /* HAVE_BLKDEV_REREAD_PART */ #else #ifdef HAVE_BDEV_CHECK_MEDIA_CHANGE static inline int zfs_check_media_change(struct block_device *bdev) { #ifdef HAVE_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK struct gendisk *gd = bdev->bd_disk; const struct block_device_operations *bdo = gd->fops; #endif if (!bdev_check_media_change(bdev)) return (0); #ifdef HAVE_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK /* * Force revalidation, to mimic the old behavior of * check_disk_change() */ if (bdo->revalidate_disk) bdo->revalidate_disk(gd); #endif return (0); } #define vdev_bdev_reread_part(bdev) zfs_check_media_change(bdev) #else /* * This is encountered if check_disk_change() and bdev_check_media_change() * are not available in the kernel - likely due to an API change that needs * to be chased down. */ #error "Unsupported kernel: no usable disk change check" #endif /* HAVE_BDEV_CHECK_MEDIA_CHANGE */ #endif /* HAVE_CHECK_DISK_CHANGE */ /* * 2.6.27 API change * The function was exported for use, prior to this it existed but the * symbol was not exported. * * 4.4.0-6.21 API change for Ubuntu * lookup_bdev() gained a second argument, FMODE_*, to check inode permissions. * * 5.11 API change * Changed to take a dev_t argument which is set on success and return a * non-zero error code on failure. */ static inline int vdev_lookup_bdev(const char *path, dev_t *dev) { #if defined(HAVE_DEVT_LOOKUP_BDEV) return (lookup_bdev(path, dev)); #elif defined(HAVE_1ARG_LOOKUP_BDEV) struct block_device *bdev = lookup_bdev(path); if (IS_ERR(bdev)) return (PTR_ERR(bdev)); *dev = bdev->bd_dev; bdput(bdev); return (0); #elif defined(HAVE_MODE_LOOKUP_BDEV) struct block_device *bdev = lookup_bdev(path, FMODE_READ); if (IS_ERR(bdev)) return (PTR_ERR(bdev)); *dev = bdev->bd_dev; bdput(bdev); return (0); #else #error "Unsupported kernel" #endif } /* * Kernels without bio_set_op_attrs use bi_rw for the bio flags. */ #if !defined(HAVE_BIO_SET_OP_ATTRS) static inline void bio_set_op_attrs(struct bio *bio, unsigned rw, unsigned flags) { bio->bi_rw |= rw | flags; } #endif /* * bio_set_flush - Set the appropriate flags in a bio to guarantee * data are on non-volatile media on completion. * * 2.6.37 - 4.8 API, * Introduce WRITE_FLUSH, WRITE_FUA, and WRITE_FLUSH_FUA flags as a * replacement for WRITE_BARRIER to allow expressing richer semantics * to the block layer. It's up to the block layer to implement the * semantics correctly. Use the WRITE_FLUSH_FUA flag combination. * * 4.8 - 4.9 API, * REQ_FLUSH was renamed to REQ_PREFLUSH. For consistency with previous * OpenZFS releases, prefer the WRITE_FLUSH_FUA flag set if it's available. * * 4.10 API, * The read/write flags and their modifiers, including WRITE_FLUSH, * WRITE_FUA and WRITE_FLUSH_FUA were removed from fs.h in * torvalds/linux@70fd7614 and replaced by direct flag modification * of the REQ_ flags in bio->bi_opf. Use REQ_PREFLUSH. */ static inline void bio_set_flush(struct bio *bio) { #if defined(HAVE_REQ_PREFLUSH) /* >= 4.10 */ bio_set_op_attrs(bio, 0, REQ_PREFLUSH); #elif defined(WRITE_FLUSH_FUA) /* >= 2.6.37 and <= 4.9 */ bio_set_op_attrs(bio, 0, WRITE_FLUSH_FUA); #else #error "Allowing the build will cause bio_set_flush requests to be ignored." #endif } /* * 4.8 API, * REQ_OP_FLUSH * * 4.8-rc0 - 4.8-rc1, * REQ_PREFLUSH * * 2.6.36 - 4.7 API, * REQ_FLUSH * * in all cases but may have a performance impact for some kernels. It * has the advantage of minimizing kernel specific changes in the zvol code. * */ static inline boolean_t bio_is_flush(struct bio *bio) { #if defined(HAVE_REQ_OP_FLUSH) && defined(HAVE_BIO_BI_OPF) return ((bio_op(bio) == REQ_OP_FLUSH) || (bio->bi_opf & REQ_PREFLUSH)); #elif defined(HAVE_REQ_PREFLUSH) && defined(HAVE_BIO_BI_OPF) return (bio->bi_opf & REQ_PREFLUSH); #elif defined(HAVE_REQ_PREFLUSH) && !defined(HAVE_BIO_BI_OPF) return (bio->bi_rw & REQ_PREFLUSH); #elif defined(HAVE_REQ_FLUSH) return (bio->bi_rw & REQ_FLUSH); #else #error "Unsupported kernel" #endif } /* * 4.8 API, * REQ_FUA flag moved to bio->bi_opf * * 2.6.x - 4.7 API, * REQ_FUA */ static inline boolean_t bio_is_fua(struct bio *bio) { #if defined(HAVE_BIO_BI_OPF) return (bio->bi_opf & REQ_FUA); #elif defined(REQ_FUA) return (bio->bi_rw & REQ_FUA); #else #error "Allowing the build will cause fua requests to be ignored." #endif } /* * 4.8 API, * REQ_OP_DISCARD * * 2.6.36 - 4.7 API, * REQ_DISCARD * * In all cases the normal I/O path is used for discards. The only * difference is how the kernel tags individual I/Os as discards. */ static inline boolean_t bio_is_discard(struct bio *bio) { #if defined(HAVE_REQ_OP_DISCARD) return (bio_op(bio) == REQ_OP_DISCARD); #elif defined(HAVE_REQ_DISCARD) return (bio->bi_rw & REQ_DISCARD); #else #error "Unsupported kernel" #endif } /* * 4.8 API, * REQ_OP_SECURE_ERASE * * 2.6.36 - 4.7 API, * REQ_SECURE */ static inline boolean_t bio_is_secure_erase(struct bio *bio) { #if defined(HAVE_REQ_OP_SECURE_ERASE) return (bio_op(bio) == REQ_OP_SECURE_ERASE); #elif defined(REQ_SECURE) return (bio->bi_rw & REQ_SECURE); #else return (0); #endif } /* * 2.6.33 API change * Discard granularity and alignment restrictions may now be set. For * older kernels which do not support this it is safe to skip it. */ static inline void blk_queue_discard_granularity(struct request_queue *q, unsigned int dg) { q->limits.discard_granularity = dg; } /* * 5.19 API, * bdev_max_discard_sectors() * * 2.6.32 API, * blk_queue_discard() */ static inline boolean_t bdev_discard_supported(struct block_device *bdev) { #if defined(HAVE_BDEV_MAX_DISCARD_SECTORS) return (!!bdev_max_discard_sectors(bdev)); #elif defined(HAVE_BLK_QUEUE_DISCARD) return (!!blk_queue_discard(bdev_get_queue(bdev))); #else #error "Unsupported kernel" #endif } /* * 5.19 API, * bdev_max_secure_erase_sectors() * * 4.8 API, * blk_queue_secure_erase() * * 2.6.36 - 4.7 API, * blk_queue_secdiscard() */ static inline boolean_t bdev_secure_discard_supported(struct block_device *bdev) { #if defined(HAVE_BDEV_MAX_SECURE_ERASE_SECTORS) return (!!bdev_max_secure_erase_sectors(bdev)); #elif defined(HAVE_BLK_QUEUE_SECURE_ERASE) return (!!blk_queue_secure_erase(bdev_get_queue(bdev))); #elif defined(HAVE_BLK_QUEUE_SECDISCARD) return (!!blk_queue_secdiscard(bdev_get_queue(bdev))); #else #error "Unsupported kernel" #endif } /* * A common holder for vdev_bdev_open() is used to relax the exclusive open * semantics slightly. Internal vdev disk callers may pass VDEV_HOLDER to * allow them to open the device multiple times. Other kernel callers and * user space processes which don't pass this value will get EBUSY. This is * currently required for the correct operation of hot spares. */ #define VDEV_HOLDER ((void *)0x2401de7) static inline unsigned long blk_generic_start_io_acct(struct request_queue *q __attribute__((unused)), struct gendisk *disk __attribute__((unused)), int rw __attribute__((unused)), struct bio *bio) { #if defined(HAVE_BDEV_IO_ACCT) return (bdev_start_io_acct(bio->bi_bdev, bio_sectors(bio), bio_op(bio), jiffies)); #elif defined(HAVE_DISK_IO_ACCT) return (disk_start_io_acct(disk, bio_sectors(bio), bio_op(bio))); #elif defined(HAVE_BIO_IO_ACCT) return (bio_start_io_acct(bio)); #elif defined(HAVE_GENERIC_IO_ACCT_3ARG) unsigned long start_time = jiffies; generic_start_io_acct(rw, bio_sectors(bio), &disk->part0); return (start_time); #elif defined(HAVE_GENERIC_IO_ACCT_4ARG) unsigned long start_time = jiffies; generic_start_io_acct(q, rw, bio_sectors(bio), &disk->part0); return (start_time); #else /* Unsupported */ return (0); #endif } static inline void blk_generic_end_io_acct(struct request_queue *q __attribute__((unused)), struct gendisk *disk __attribute__((unused)), int rw __attribute__((unused)), struct bio *bio, unsigned long start_time) { #if defined(HAVE_BDEV_IO_ACCT) bdev_end_io_acct(bio->bi_bdev, bio_op(bio), start_time); #elif defined(HAVE_DISK_IO_ACCT) disk_end_io_acct(disk, bio_op(bio), start_time); #elif defined(HAVE_BIO_IO_ACCT) bio_end_io_acct(bio, start_time); #elif defined(HAVE_GENERIC_IO_ACCT_3ARG) generic_end_io_acct(rw, &disk->part0, start_time); #elif defined(HAVE_GENERIC_IO_ACCT_4ARG) generic_end_io_acct(q, rw, &disk->part0, start_time); #endif } #ifndef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS static inline struct request_queue * blk_generic_alloc_queue(make_request_fn make_request, int node_id) { #if defined(HAVE_BLK_ALLOC_QUEUE_REQUEST_FN) return (blk_alloc_queue(make_request, node_id)); #elif defined(HAVE_BLK_ALLOC_QUEUE_REQUEST_FN_RH) return (blk_alloc_queue_rh(make_request, node_id)); #else struct request_queue *q = blk_alloc_queue(GFP_KERNEL); if (q != NULL) blk_queue_make_request(q, make_request); return (q); #endif } #endif /* !HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS */ /* * All the io_*() helper functions below can operate on a bio, or a rq, but * not both. The older submit_bio() codepath will pass a bio, and the * newer blk-mq codepath will pass a rq. */ static inline int io_data_dir(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) { if (op_is_write(req_op(rq))) { return (WRITE); } else { return (READ); } } #else ASSERT3P(rq, ==, NULL); #endif return (bio_data_dir(bio)); } static inline int io_is_flush(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (req_op(rq) == REQ_OP_FLUSH); #else ASSERT3P(rq, ==, NULL); #endif return (bio_is_flush(bio)); } static inline int io_is_discard(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (req_op(rq) == REQ_OP_DISCARD); #else ASSERT3P(rq, ==, NULL); #endif return (bio_is_discard(bio)); } static inline int io_is_secure_erase(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (req_op(rq) == REQ_OP_SECURE_ERASE); #else ASSERT3P(rq, ==, NULL); #endif return (bio_is_secure_erase(bio)); } static inline int io_is_fua(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (rq->cmd_flags & REQ_FUA); #else ASSERT3P(rq, ==, NULL); #endif return (bio_is_fua(bio)); } static inline uint64_t io_offset(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (blk_rq_pos(rq) << 9); #else ASSERT3P(rq, ==, NULL); #endif return (BIO_BI_SECTOR(bio) << 9); } static inline uint64_t io_size(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (blk_rq_bytes(rq)); #else ASSERT3P(rq, ==, NULL); #endif return (BIO_BI_SIZE(bio)); } static inline int io_has_data(struct bio *bio, struct request *rq) { #ifdef HAVE_BLK_MQ if (rq != NULL) return (bio_has_data(rq->bio)); #else ASSERT3P(rq, ==, NULL); #endif return (bio_has_data(bio)); } #endif /* _ZFS_BLKDEV_H */ diff --git a/include/os/linux/spl/sys/misc.h b/include/os/linux/spl/sys/misc.h new file mode 100644 index 000000000000..299fe9c1ab07 --- /dev/null +++ b/include/os/linux/spl/sys/misc.h @@ -0,0 +1,29 @@ +/* + * CDDL HEADER START + * + * The contents of this file are subject to the terms of the + * Common Development and Distribution License (the "License"). + * You may not use this file except in compliance with the License. + * + * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE + * or https://opensource.org/licenses/CDDL-1.0. + * See the License for the specific language governing permissions + * and limitations under the License. + * + * When distributing Covered Code, include this CDDL HEADER in each + * file and include the License file at usr/src/OPENSOLARIS.LICENSE. + * If applicable, add the following below this CDDL HEADER, with the + * fields enclosed by brackets "[]" replaced with your own identifying + * information: Portions Copyright [yyyy] [name of copyright owner] + * + * CDDL HEADER END + */ + +#ifndef _OS_LINUX_SPL_MISC_H +#define _OS_LINUX_SPL_MISC_H + +#include + +extern void spl_signal_kobj_evt(struct block_device *bdev); + +#endif diff --git a/include/sys/spa.h b/include/sys/spa.h index 76fae788bc1f..3e68cb8c6511 100644 --- a/include/sys/spa.h +++ b/include/sys/spa.h @@ -1,1232 +1,1232 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2021 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 Joyent, Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Allan Jude * Copyright (c) 2019, Klara Inc. * Copyright (c) 2019, Datto Inc. */ #ifndef _SYS_SPA_H #define _SYS_SPA_H #include #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Forward references that lots of things need. */ typedef struct spa spa_t; typedef struct vdev vdev_t; typedef struct metaslab metaslab_t; typedef struct metaslab_group metaslab_group_t; typedef struct metaslab_class metaslab_class_t; typedef struct zio zio_t; typedef struct zilog zilog_t; typedef struct spa_aux_vdev spa_aux_vdev_t; typedef struct ddt ddt_t; typedef struct ddt_entry ddt_entry_t; typedef struct zbookmark_phys zbookmark_phys_t; struct bpobj; struct bplist; struct dsl_pool; struct dsl_dataset; struct dsl_crypto_params; /* * Alignment Shift (ashift) is an immutable, internal top-level vdev property * which can only be set at vdev creation time. Physical writes are always done * according to it, which makes 2^ashift the smallest possible IO on a vdev. * * We currently allow values ranging from 512 bytes (2^9 = 512) to 64 KiB * (2^16 = 65,536). */ #define ASHIFT_MIN 9 #define ASHIFT_MAX 16 /* * Size of block to hold the configuration data (a packed nvlist) */ #define SPA_CONFIG_BLOCKSIZE (1ULL << 14) /* * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB. * The ASIZE encoding should be at least 64 times larger (6 more bits) * to support up to 4-way RAID-Z mirror mode with worst-case gang block * overhead, three DVAs per bp, plus one more bit in case we do anything * else that expands the ASIZE. */ #define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */ #define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */ #define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */ #define SPA_COMPRESSBITS 7 #define SPA_VDEVBITS 24 #define SPA_COMPRESSMASK ((1U << SPA_COMPRESSBITS) - 1) /* * All SPA data is represented by 128-bit data virtual addresses (DVAs). * The members of the dva_t should be considered opaque outside the SPA. */ typedef struct dva { uint64_t dva_word[2]; } dva_t; /* * Some checksums/hashes need a 256-bit initialization salt. This salt is kept * secret and is suitable for use in MAC algorithms as the key. */ typedef struct zio_cksum_salt { uint8_t zcs_bytes[32]; } zio_cksum_salt_t; /* * Each block is described by its DVAs, time of birth, checksum, etc. * The word-by-word, bit-by-bit layout of the blkptr is as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | pad | vdev1 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 1 |G| offset1 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 2 | pad | vdev2 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 3 |G| offset2 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 4 | pad | vdev3 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 5 |G| offset3 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | cksum |E| comp| PSIZE | LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 8 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 9 | physical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | fill count | * +-------+-------+-------+-------+-------+-------+-------+-------+ * c | checksum[0] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * d | checksum[1] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * e | checksum[2] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * f | checksum[3] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * vdev virtual device ID * offset offset into virtual device * LSIZE logical size * PSIZE physical size (after compression) * ASIZE allocated size (including RAID-Z parity and gang block headers) * GRID RAID-Z layout information (reserved for future use) * cksum checksum function * comp compression function * G gang block indicator * B byteorder (endianness) * D dedup * X encryption * E blkptr_t contains embedded data (see below) * lvl level of indirection * type DMU object type * phys birth txg when dva[0] was written; zero if same as logical birth txg * note that typically all the dva's would be written in this * txg, but they could be different if they were moved by * device removal. * log. birth transaction group in which the block was logically born * fill count number of non-zero blocks under this bp * checksum[4] 256-bit checksum of the data this bp describes */ /* * The blkptr_t's of encrypted blocks also need to store the encryption * parameters so that the block can be decrypted. This layout is as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | vdev1 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 1 |G| offset1 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 2 | vdev2 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 3 |G| offset2 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 4 | salt | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 5 | IV1 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | cksum |E| comp| PSIZE | LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 8 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 9 | physical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | IV2 | fill count | * +-------+-------+-------+-------+-------+-------+-------+-------+ * c | checksum[0] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * d | checksum[1] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * e | MAC[0] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * f | MAC[1] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * salt Salt for generating encryption keys * IV1 First 64 bits of encryption IV * X Block requires encryption handling (set to 1) * E blkptr_t contains embedded data (set to 0, see below) * fill count number of non-zero blocks under this bp (truncated to 32 bits) * IV2 Last 32 bits of encryption IV * checksum[2] 128-bit checksum of the data this bp describes * MAC[2] 128-bit message authentication code for this data * * The X bit being set indicates that this block is one of 3 types. If this is * a level 0 block with an encrypted object type, the block is encrypted * (see BP_IS_ENCRYPTED()). If this is a level 0 block with an unencrypted * object type, this block is authenticated with an HMAC (see * BP_IS_AUTHENTICATED()). Otherwise (if level > 0), this bp will use the MAC * words to store a checksum-of-MACs from the level below (see * BP_HAS_INDIRECT_MAC_CKSUM()). For convenience in the code, BP_IS_PROTECTED() * refers to both encrypted and authenticated blocks and BP_USES_CRYPT() * refers to any of these 3 kinds of blocks. * * The additional encryption parameters are the salt, IV, and MAC which are * explained in greater detail in the block comment at the top of zio_crypt.c. * The MAC occupies half of the checksum space since it serves a very similar * purpose: to prevent data corruption on disk. The only functional difference * is that the checksum is used to detect on-disk corruption whether or not the * encryption key is loaded and the MAC provides additional protection against * malicious disk tampering. We use the 3rd DVA to store the salt and first * 64 bits of the IV. As a result encrypted blocks can only have 2 copies * maximum instead of the normal 3. The last 32 bits of the IV are stored in * the upper bits of what is usually the fill count. Note that only blocks at * level 0 or -2 are ever encrypted, which allows us to guarantee that these * 32 bits are not trampled over by other code (see zio_crypt.c for details). * The salt and IV are not used for authenticated bps or bps with an indirect * MAC checksum, so these blocks can utilize all 3 DVAs and the full 64 bits * for the fill count. */ /* * "Embedded" blkptr_t's don't actually point to a block, instead they * have a data payload embedded in the blkptr_t itself. See the comment * in blkptr.c for more details. * * The blkptr_t is laid out as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | payload | * 1 | payload | * 2 | payload | * 3 | payload | * 4 | payload | * 5 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | etype |E| comp| PSIZE| LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | payload | * 8 | payload | * 9 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | payload | * c | payload | * d | payload | * e | payload | * f | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * payload contains the embedded data * B (byteorder) byteorder (endianness) * D (dedup) padding (set to zero) * X encryption (set to zero) * E (embedded) set to one * lvl indirection level * type DMU object type * etype how to interpret embedded data (BP_EMBEDDED_TYPE_*) * comp compression function of payload * PSIZE size of payload after compression, in bytes * LSIZE logical size of payload, in bytes * note that 25 bits is enough to store the largest * "normal" BP's LSIZE (2^16 * 2^9) in bytes * log. birth transaction group in which the block was logically born * * Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded * bp's they are stored in units of SPA_MINBLOCKSHIFT. * Generally, the generic BP_GET_*() macros can be used on embedded BP's. * The B, D, X, lvl, type, and comp fields are stored the same as with normal * BP's so the BP_SET_* macros can be used with them. etype, PSIZE, LSIZE must * be set with the BPE_SET_* macros. BP_SET_EMBEDDED() should be called before * other macros, as they assert that they are only used on BP's of the correct * "embedded-ness". Encrypted blkptr_t's cannot be embedded because they use * the payload space for encryption parameters (see the comment above on * how encryption parameters are stored). */ #define BPE_GET_ETYPE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET((bp)->blk_prop, 40, 8)) #define BPE_SET_ETYPE(bp, t) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, t); \ } while (0) #define BPE_GET_LSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1)) #define BPE_SET_LSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \ } while (0) #define BPE_GET_PSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1)) #define BPE_SET_PSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \ } while (0) typedef enum bp_embedded_type { BP_EMBEDDED_TYPE_DATA, BP_EMBEDDED_TYPE_RESERVED, /* Reserved for Delphix byteswap feature. */ BP_EMBEDDED_TYPE_REDACTED, NUM_BP_EMBEDDED_TYPES } bp_embedded_type_t; #define BPE_NUM_WORDS 14 #define BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t)) #define BPE_IS_PAYLOADWORD(bp, wp) \ ((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth) #define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */ #define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */ #define SPA_SYNC_MIN_VDEVS 3 /* min vdevs to update during sync */ /* * A block is a hole when it has either 1) never been written to, or * 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads * without physically allocating disk space. Holes are represented in the * blkptr_t structure by zeroed blk_dva. Correct checking for holes is * done through the BP_IS_HOLE macro. For holes, the logical size, level, * DMU object type, and birth times are all also stored for holes that * were written to at some point (i.e. were punched after having been filled). */ typedef struct blkptr { dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */ uint64_t blk_prop; /* size, compression, type, etc */ uint64_t blk_pad[2]; /* Extra space for the future */ uint64_t blk_phys_birth; /* txg when block was allocated */ uint64_t blk_birth; /* transaction group at birth */ uint64_t blk_fill; /* fill count */ zio_cksum_t blk_cksum; /* 256-bit checksum */ } blkptr_t; /* * Macros to get and set fields in a bp or DVA. */ /* * Note, for gang blocks, DVA_GET_ASIZE() is the total space allocated for * this gang DVA including its children BP's. The space allocated at this * DVA's vdev/offset is vdev_gang_header_asize(vdev). */ #define DVA_GET_ASIZE(dva) \ BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_ASIZE(dva, x) \ BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \ SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8) #define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x) #define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, SPA_VDEVBITS) #define DVA_SET_VDEV(dva, x) \ BF64_SET((dva)->dva_word[0], 32, SPA_VDEVBITS, x) #define DVA_GET_OFFSET(dva) \ BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_OFFSET(dva, x) \ BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1) #define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x) #define BP_GET_LSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? \ (BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \ BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_LSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ } while (0) #define BP_GET_PSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_PSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ } while (0) #define BP_GET_COMPRESS(bp) \ BF64_GET((bp)->blk_prop, 32, SPA_COMPRESSBITS) #define BP_SET_COMPRESS(bp, x) \ BF64_SET((bp)->blk_prop, 32, SPA_COMPRESSBITS, x) #define BP_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1) #define BP_SET_EMBEDDED(bp, x) BF64_SET((bp)->blk_prop, 39, 1, x) #define BP_GET_CHECKSUM(bp) \ (BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \ BF64_GET((bp)->blk_prop, 40, 8)) #define BP_SET_CHECKSUM(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, x); \ } while (0) #define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8) #define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x) #define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5) #define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x) /* encrypted, authenticated, and MAC cksum bps use the same bit */ #define BP_USES_CRYPT(bp) BF64_GET((bp)->blk_prop, 61, 1) #define BP_SET_CRYPT(bp, x) BF64_SET((bp)->blk_prop, 61, 1, x) #define BP_IS_ENCRYPTED(bp) \ (BP_USES_CRYPT(bp) && \ BP_GET_LEVEL(bp) <= 0 && \ DMU_OT_IS_ENCRYPTED(BP_GET_TYPE(bp))) #define BP_IS_AUTHENTICATED(bp) \ (BP_USES_CRYPT(bp) && \ BP_GET_LEVEL(bp) <= 0 && \ !DMU_OT_IS_ENCRYPTED(BP_GET_TYPE(bp))) #define BP_HAS_INDIRECT_MAC_CKSUM(bp) \ (BP_USES_CRYPT(bp) && BP_GET_LEVEL(bp) > 0) #define BP_IS_PROTECTED(bp) \ (BP_IS_ENCRYPTED(bp) || BP_IS_AUTHENTICATED(bp)) #define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1) #define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x) #define BP_GET_BYTEORDER(bp) BF64_GET((bp)->blk_prop, 63, 1) #define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x) #define BP_GET_FREE(bp) BF64_GET((bp)->blk_fill, 0, 1) #define BP_SET_FREE(bp, x) BF64_SET((bp)->blk_fill, 0, 1, x) #define BP_PHYSICAL_BIRTH(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth) #define BP_SET_BIRTH(bp, logical, physical) \ { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ (bp)->blk_birth = (logical); \ (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \ } #define BP_GET_FILL(bp) \ ((BP_IS_ENCRYPTED(bp)) ? BF64_GET((bp)->blk_fill, 0, 32) : \ ((BP_IS_EMBEDDED(bp)) ? 1 : (bp)->blk_fill)) #define BP_SET_FILL(bp, fill) \ { \ if (BP_IS_ENCRYPTED(bp)) \ BF64_SET((bp)->blk_fill, 0, 32, fill); \ else \ (bp)->blk_fill = fill; \ } #define BP_GET_IV2(bp) \ (ASSERT(BP_IS_ENCRYPTED(bp)), \ BF64_GET((bp)->blk_fill, 32, 32)) #define BP_SET_IV2(bp, iv2) \ { \ ASSERT(BP_IS_ENCRYPTED(bp)); \ BF64_SET((bp)->blk_fill, 32, 32, iv2); \ } #define BP_IS_METADATA(bp) \ (BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) #define BP_GET_ASIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ (DVA_GET_ASIZE(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp))) #define BP_GET_UCSIZE(bp) \ (BP_IS_METADATA(bp) ? BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)) #define BP_GET_NDVAS(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ !!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ (!!DVA_GET_ASIZE(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp))) #define BP_COUNT_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (DVA_GET_GANG(&(bp)->blk_dva[0]) + \ DVA_GET_GANG(&(bp)->blk_dva[1]) + \ (DVA_GET_GANG(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp)))) #define DVA_EQUAL(dva1, dva2) \ ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \ (dva1)->dva_word[0] == (dva2)->dva_word[0]) #define BP_EQUAL(bp1, bp2) \ (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \ (bp1)->blk_birth == (bp2)->blk_birth && \ DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \ DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \ DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2])) #define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0) #define BP_IDENTITY(bp) (ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0]) #define BP_IS_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp))) #define DVA_IS_EMPTY(dva) ((dva)->dva_word[0] == 0ULL && \ (dva)->dva_word[1] == 0ULL) #define BP_IS_HOLE(bp) \ (!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp))) #define BP_SET_REDACTED(bp) \ { \ BP_SET_EMBEDDED(bp, B_TRUE); \ BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_REDACTED); \ } #define BP_IS_REDACTED(bp) \ (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_REDACTED) /* BP_IS_RAIDZ(bp) assumes no block compression */ #define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \ BP_GET_PSIZE(bp)) #define BP_ZERO(bp) \ { \ (bp)->blk_dva[0].dva_word[0] = 0; \ (bp)->blk_dva[0].dva_word[1] = 0; \ (bp)->blk_dva[1].dva_word[0] = 0; \ (bp)->blk_dva[1].dva_word[1] = 0; \ (bp)->blk_dva[2].dva_word[0] = 0; \ (bp)->blk_dva[2].dva_word[1] = 0; \ (bp)->blk_prop = 0; \ (bp)->blk_pad[0] = 0; \ (bp)->blk_pad[1] = 0; \ (bp)->blk_phys_birth = 0; \ (bp)->blk_birth = 0; \ (bp)->blk_fill = 0; \ ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \ } #ifdef _ZFS_BIG_ENDIAN #define ZFS_HOST_BYTEORDER (0ULL) #else #define ZFS_HOST_BYTEORDER (1ULL) #endif #define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER) #define BP_SPRINTF_LEN 400 /* * This macro allows code sharing between zfs, libzpool, and mdb. * 'func' is either snprintf() or mdb_snprintf(). * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line. */ #define SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \ { \ static const char *const copyname[] = \ { "zero", "single", "double", "triple" }; \ int len = 0; \ int copies = 0; \ const char *crypt_type; \ if (bp != NULL) { \ if (BP_IS_ENCRYPTED(bp)) { \ crypt_type = "encrypted"; \ /* LINTED E_SUSPICIOUS_COMPARISON */ \ } else if (BP_IS_AUTHENTICATED(bp)) { \ crypt_type = "authenticated"; \ } else if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) { \ crypt_type = "indirect-MAC"; \ } else { \ crypt_type = "unencrypted"; \ } \ } \ if (bp == NULL) { \ len += func(buf + len, size - len, ""); \ } else if (BP_IS_HOLE(bp)) { \ len += func(buf + len, size - len, \ "HOLE [L%llu %s] " \ "size=%llxL birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else if (BP_IS_EMBEDDED(bp)) { \ len = func(buf + len, size - len, \ "EMBEDDED [L%llu %s] et=%u %s " \ "size=%llxL/%llxP birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (int)BPE_GET_ETYPE(bp), \ compress, \ (u_longlong_t)BPE_GET_LSIZE(bp), \ (u_longlong_t)BPE_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else if (BP_IS_REDACTED(bp)) { \ len += func(buf + len, size - len, \ "REDACTED [L%llu %s] size=%llxL birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else { \ for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \ const dva_t *dva = &bp->blk_dva[d]; \ if (DVA_IS_VALID(dva)) \ copies++; \ len += func(buf + len, size - len, \ "DVA[%d]=<%llu:%llx:%llx>%c", d, \ (u_longlong_t)DVA_GET_VDEV(dva), \ (u_longlong_t)DVA_GET_OFFSET(dva), \ (u_longlong_t)DVA_GET_ASIZE(dva), \ ws); \ } \ if (BP_IS_ENCRYPTED(bp)) { \ len += func(buf + len, size - len, \ "salt=%llx iv=%llx:%llx%c", \ (u_longlong_t)bp->blk_dva[2].dva_word[0], \ (u_longlong_t)bp->blk_dva[2].dva_word[1], \ (u_longlong_t)BP_GET_IV2(bp), \ ws); \ } \ if (BP_IS_GANG(bp) && \ DVA_GET_ASIZE(&bp->blk_dva[2]) <= \ DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \ copies--; \ len += func(buf + len, size - len, \ "[L%llu %s] %s %s %s %s %s %s %s%c" \ "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \ "cksum=%llx:%llx:%llx:%llx", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ checksum, \ compress, \ crypt_type, \ BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \ BP_IS_GANG(bp) ? "gang" : "contiguous", \ BP_GET_DEDUP(bp) ? "dedup" : "unique", \ copyname[copies], \ ws, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)BP_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth, \ (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \ (u_longlong_t)BP_GET_FILL(bp), \ ws, \ (u_longlong_t)bp->blk_cksum.zc_word[0], \ (u_longlong_t)bp->blk_cksum.zc_word[1], \ (u_longlong_t)bp->blk_cksum.zc_word[2], \ (u_longlong_t)bp->blk_cksum.zc_word[3]); \ } \ ASSERT(len < size); \ } #define BP_GET_BUFC_TYPE(bp) \ (BP_IS_METADATA(bp) ? ARC_BUFC_METADATA : ARC_BUFC_DATA) typedef enum spa_import_type { SPA_IMPORT_EXISTING, SPA_IMPORT_ASSEMBLE } spa_import_type_t; typedef enum spa_mode { SPA_MODE_UNINIT = 0, SPA_MODE_READ = 1, SPA_MODE_WRITE = 2, } spa_mode_t; /* * Send TRIM commands in-line during normal pool operation while deleting. * OFF: no * ON: yes * NB: IN_FREEBSD_BASE is defined within the FreeBSD sources. */ typedef enum { SPA_AUTOTRIM_OFF = 0, /* default */ SPA_AUTOTRIM_ON, #ifdef IN_FREEBSD_BASE SPA_AUTOTRIM_DEFAULT = SPA_AUTOTRIM_ON, #else SPA_AUTOTRIM_DEFAULT = SPA_AUTOTRIM_OFF, #endif } spa_autotrim_t; /* * Reason TRIM command was issued, used internally for accounting purposes. */ typedef enum trim_type { TRIM_TYPE_MANUAL = 0, TRIM_TYPE_AUTO = 1, TRIM_TYPE_SIMPLE = 2 } trim_type_t; /* state manipulation functions */ extern int spa_open(const char *pool, spa_t **, const void *tag); extern int spa_open_rewind(const char *pool, spa_t **, const void *tag, nvlist_t *policy, nvlist_t **config); extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot, size_t buflen); extern int spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *zplprops, struct dsl_crypto_params *dcp); extern int spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags); extern nvlist_t *spa_tryimport(nvlist_t *tryconfig); extern int spa_destroy(const char *pool); extern int spa_checkpoint(const char *pool); extern int spa_checkpoint_discard(const char *pool); extern int spa_export(const char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce); extern int spa_reset(const char *pool); extern void spa_async_request(spa_t *spa, int flag); extern void spa_async_unrequest(spa_t *spa, int flag); extern void spa_async_suspend(spa_t *spa); extern void spa_async_resume(spa_t *spa); extern int spa_async_tasks(spa_t *spa); extern spa_t *spa_inject_addref(char *pool); extern void spa_inject_delref(spa_t *spa); extern void spa_scan_stat_init(spa_t *spa); extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps); extern int bpobj_enqueue_alloc_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx); extern int bpobj_enqueue_free_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx); #define SPA_ASYNC_CONFIG_UPDATE 0x01 #define SPA_ASYNC_REMOVE 0x02 #define SPA_ASYNC_PROBE 0x04 #define SPA_ASYNC_RESILVER_DONE 0x08 #define SPA_ASYNC_RESILVER 0x10 #define SPA_ASYNC_AUTOEXPAND 0x20 #define SPA_ASYNC_REMOVE_DONE 0x40 #define SPA_ASYNC_REMOVE_STOP 0x80 #define SPA_ASYNC_INITIALIZE_RESTART 0x100 #define SPA_ASYNC_TRIM_RESTART 0x200 #define SPA_ASYNC_AUTOTRIM_RESTART 0x400 #define SPA_ASYNC_L2CACHE_REBUILD 0x800 #define SPA_ASYNC_L2CACHE_TRIM 0x1000 #define SPA_ASYNC_REBUILD_DONE 0x2000 /* device manipulation */ extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot); extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing, int rebuild); extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done); extern int spa_vdev_alloc(spa_t *spa, uint64_t guid); extern int spa_vdev_noalloc(spa_t *spa, uint64_t guid); extern boolean_t spa_vdev_remove_active(spa_t *spa); extern int spa_vdev_initialize(spa_t *spa, nvlist_t *nv, uint64_t cmd_type, nvlist_t *vdev_errlist); extern int spa_vdev_trim(spa_t *spa, nvlist_t *nv, uint64_t cmd_type, uint64_t rate, boolean_t partial, boolean_t secure, nvlist_t *vdev_errlist); extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath); extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru); extern int spa_vdev_split_mirror(spa_t *spa, const char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp); /* spare state (which is global across all pools) */ extern void spa_spare_add(vdev_t *vd); extern void spa_spare_remove(vdev_t *vd); extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt); extern void spa_spare_activate(vdev_t *vd); /* L2ARC state (which is global across all pools) */ extern void spa_l2cache_add(vdev_t *vd); extern void spa_l2cache_remove(vdev_t *vd); extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool); extern void spa_l2cache_activate(vdev_t *vd); extern void spa_l2cache_drop(spa_t *spa); /* scanning */ extern int spa_scan(spa_t *spa, pool_scan_func_t func); extern int spa_scan_stop(spa_t *spa); extern int spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t flag); /* spa syncing */ extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */ extern void spa_sync_allpools(void); extern uint_t zfs_sync_pass_deferred_free; /* spa namespace global mutex */ extern kmutex_t spa_namespace_lock; /* * SPA configuration functions in spa_config.c */ #define SPA_CONFIG_UPDATE_POOL 0 #define SPA_CONFIG_UPDATE_VDEVS 1 -extern void spa_write_cachefile(spa_t *, boolean_t, boolean_t); +extern void spa_write_cachefile(spa_t *, boolean_t, boolean_t, boolean_t); extern void spa_config_load(void); extern nvlist_t *spa_all_configs(uint64_t *); extern void spa_config_set(spa_t *spa, nvlist_t *config); extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats); extern void spa_config_update(spa_t *spa, int what); extern int spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int atype); /* * Miscellaneous SPA routines in spa_misc.c */ /* Namespace manipulation */ extern spa_t *spa_lookup(const char *name); extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot); extern void spa_remove(spa_t *spa); extern spa_t *spa_next(spa_t *prev); /* Refcount functions */ extern void spa_open_ref(spa_t *spa, const void *tag); extern void spa_close(spa_t *spa, const void *tag); extern void spa_async_close(spa_t *spa, const void *tag); extern boolean_t spa_refcount_zero(spa_t *spa); #define SCL_NONE 0x00 #define SCL_CONFIG 0x01 #define SCL_STATE 0x02 #define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */ #define SCL_ALLOC 0x08 #define SCL_ZIO 0x10 #define SCL_FREE 0x20 #define SCL_VDEV 0x40 #define SCL_LOCKS 7 #define SCL_ALL ((1 << SCL_LOCKS) - 1) #define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO) /* Historical pool statistics */ typedef struct spa_history_kstat { kmutex_t lock; uint64_t count; uint64_t size; kstat_t *kstat; void *priv; list_t list; } spa_history_kstat_t; typedef struct spa_history_list { uint64_t size; procfs_list_t procfs_list; } spa_history_list_t; typedef struct spa_stats { spa_history_list_t read_history; spa_history_list_t txg_history; spa_history_kstat_t tx_assign_histogram; spa_history_list_t mmp_history; spa_history_kstat_t state; /* pool state */ spa_history_kstat_t guid; /* pool guid */ spa_history_kstat_t iostats; } spa_stats_t; typedef enum txg_state { TXG_STATE_BIRTH = 0, TXG_STATE_OPEN = 1, TXG_STATE_QUIESCED = 2, TXG_STATE_WAIT_FOR_SYNC = 3, TXG_STATE_SYNCED = 4, TXG_STATE_COMMITTED = 5, } txg_state_t; typedef struct txg_stat { vdev_stat_t vs1; vdev_stat_t vs2; uint64_t txg; uint64_t ndirty; } txg_stat_t; /* Assorted pool IO kstats */ typedef struct spa_iostats { kstat_named_t trim_extents_written; kstat_named_t trim_bytes_written; kstat_named_t trim_extents_skipped; kstat_named_t trim_bytes_skipped; kstat_named_t trim_extents_failed; kstat_named_t trim_bytes_failed; kstat_named_t autotrim_extents_written; kstat_named_t autotrim_bytes_written; kstat_named_t autotrim_extents_skipped; kstat_named_t autotrim_bytes_skipped; kstat_named_t autotrim_extents_failed; kstat_named_t autotrim_bytes_failed; kstat_named_t simple_trim_extents_written; kstat_named_t simple_trim_bytes_written; kstat_named_t simple_trim_extents_skipped; kstat_named_t simple_trim_bytes_skipped; kstat_named_t simple_trim_extents_failed; kstat_named_t simple_trim_bytes_failed; } spa_iostats_t; extern void spa_stats_init(spa_t *spa); extern void spa_stats_destroy(spa_t *spa); extern void spa_read_history_add(spa_t *spa, const zbookmark_phys_t *zb, uint32_t aflags); extern void spa_txg_history_add(spa_t *spa, uint64_t txg, hrtime_t birth_time); extern int spa_txg_history_set(spa_t *spa, uint64_t txg, txg_state_t completed_state, hrtime_t completed_time); extern txg_stat_t *spa_txg_history_init_io(spa_t *, uint64_t, struct dsl_pool *); extern void spa_txg_history_fini_io(spa_t *, txg_stat_t *); extern void spa_tx_assign_add_nsecs(spa_t *spa, uint64_t nsecs); extern int spa_mmp_history_set_skip(spa_t *spa, uint64_t mmp_kstat_id); extern int spa_mmp_history_set(spa_t *spa, uint64_t mmp_kstat_id, int io_error, hrtime_t duration); extern void spa_mmp_history_add(spa_t *spa, uint64_t txg, uint64_t timestamp, uint64_t mmp_delay, vdev_t *vd, int label, uint64_t mmp_kstat_id, int error); extern void spa_iostats_trim_add(spa_t *spa, trim_type_t type, uint64_t extents_written, uint64_t bytes_written, uint64_t extents_skipped, uint64_t bytes_skipped, uint64_t extents_failed, uint64_t bytes_failed); extern void spa_import_progress_add(spa_t *spa); extern void spa_import_progress_remove(uint64_t spa_guid); extern int spa_import_progress_set_mmp_check(uint64_t pool_guid, uint64_t mmp_sec_remaining); extern int spa_import_progress_set_max_txg(uint64_t pool_guid, uint64_t max_txg); extern int spa_import_progress_set_state(uint64_t pool_guid, spa_load_state_t spa_load_state); /* Pool configuration locks */ extern int spa_config_tryenter(spa_t *spa, int locks, const void *tag, krw_t rw); extern void spa_config_enter(spa_t *spa, int locks, const void *tag, krw_t rw); extern void spa_config_exit(spa_t *spa, int locks, const void *tag); extern int spa_config_held(spa_t *spa, int locks, krw_t rw); /* Pool vdev add/remove lock */ extern uint64_t spa_vdev_enter(spa_t *spa); extern uint64_t spa_vdev_detach_enter(spa_t *spa, uint64_t guid); extern uint64_t spa_vdev_config_enter(spa_t *spa); extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, const char *tag); extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error); /* Pool vdev state change lock */ extern void spa_vdev_state_enter(spa_t *spa, int oplock); extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error); /* Log state */ typedef enum spa_log_state { SPA_LOG_UNKNOWN = 0, /* unknown log state */ SPA_LOG_MISSING, /* missing log(s) */ SPA_LOG_CLEAR, /* clear the log(s) */ SPA_LOG_GOOD, /* log(s) are good */ } spa_log_state_t; extern spa_log_state_t spa_get_log_state(spa_t *spa); extern void spa_set_log_state(spa_t *spa, spa_log_state_t state); extern int spa_reset_logs(spa_t *spa); /* Log claim callback */ extern void spa_claim_notify(zio_t *zio); extern void spa_deadman(void *); /* Accessor functions */ extern boolean_t spa_shutting_down(spa_t *spa); extern struct dsl_pool *spa_get_dsl(spa_t *spa); extern boolean_t spa_is_initializing(spa_t *spa); extern boolean_t spa_indirect_vdevs_loaded(spa_t *spa); extern blkptr_t *spa_get_rootblkptr(spa_t *spa); extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp); extern void spa_altroot(spa_t *, char *, size_t); extern uint32_t spa_sync_pass(spa_t *spa); extern char *spa_name(spa_t *spa); extern uint64_t spa_guid(spa_t *spa); extern uint64_t spa_load_guid(spa_t *spa); extern uint64_t spa_last_synced_txg(spa_t *spa); extern uint64_t spa_first_txg(spa_t *spa); extern uint64_t spa_syncing_txg(spa_t *spa); extern uint64_t spa_final_dirty_txg(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern pool_state_t spa_state(spa_t *spa); extern spa_load_state_t spa_load_state(spa_t *spa); extern uint64_t spa_freeze_txg(spa_t *spa); extern uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize); extern uint64_t spa_get_dspace(spa_t *spa); extern uint64_t spa_get_checkpoint_space(spa_t *spa); extern uint64_t spa_get_slop_space(spa_t *spa); extern void spa_update_dspace(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern boolean_t spa_deflate(spa_t *spa); extern metaslab_class_t *spa_normal_class(spa_t *spa); extern metaslab_class_t *spa_log_class(spa_t *spa); extern metaslab_class_t *spa_embedded_log_class(spa_t *spa); extern metaslab_class_t *spa_special_class(spa_t *spa); extern metaslab_class_t *spa_dedup_class(spa_t *spa); extern metaslab_class_t *spa_preferred_class(spa_t *spa, uint64_t size, dmu_object_type_t objtype, uint_t level, uint_t special_smallblk); extern void spa_evicting_os_register(spa_t *, objset_t *os); extern void spa_evicting_os_deregister(spa_t *, objset_t *os); extern void spa_evicting_os_wait(spa_t *spa); extern int spa_max_replication(spa_t *spa); extern int spa_prev_software_version(spa_t *spa); extern uint64_t spa_get_failmode(spa_t *spa); extern uint64_t spa_get_deadman_failmode(spa_t *spa); extern void spa_set_deadman_failmode(spa_t *spa, const char *failmode); extern boolean_t spa_suspended(spa_t *spa); extern uint64_t spa_bootfs(spa_t *spa); extern uint64_t spa_delegation(spa_t *spa); extern objset_t *spa_meta_objset(spa_t *spa); extern space_map_t *spa_syncing_log_sm(spa_t *spa); extern uint64_t spa_deadman_synctime(spa_t *spa); extern uint64_t spa_deadman_ziotime(spa_t *spa); extern uint64_t spa_dirty_data(spa_t *spa); extern spa_autotrim_t spa_get_autotrim(spa_t *spa); /* Miscellaneous support routines */ extern void spa_load_failed(spa_t *spa, const char *fmt, ...) __attribute__((format(printf, 2, 3))); extern void spa_load_note(spa_t *spa, const char *fmt, ...) __attribute__((format(printf, 2, 3))); extern void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx); extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature); extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid); extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid); extern char *spa_strdup(const char *); extern void spa_strfree(char *); extern uint64_t spa_generate_guid(spa_t *spa); extern void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp); extern void spa_freeze(spa_t *spa); extern int spa_change_guid(spa_t *spa); extern void spa_upgrade(spa_t *spa, uint64_t version); extern void spa_evict_all(void); extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t l2cache); extern boolean_t spa_has_l2cache(spa_t *, uint64_t guid); extern boolean_t spa_has_spare(spa_t *, uint64_t guid); extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva); extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp); extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp); extern boolean_t spa_has_slogs(spa_t *spa); extern boolean_t spa_is_root(spa_t *spa); extern boolean_t spa_writeable(spa_t *spa); extern boolean_t spa_has_pending_synctask(spa_t *spa); extern int spa_maxblocksize(spa_t *spa); extern int spa_maxdnodesize(spa_t *spa); extern boolean_t spa_has_checkpoint(spa_t *spa); extern boolean_t spa_importing_readonly_checkpoint(spa_t *spa); extern boolean_t spa_suspend_async_destroy(spa_t *spa); extern uint64_t spa_min_claim_txg(spa_t *spa); extern boolean_t zfs_dva_valid(spa_t *spa, const dva_t *dva, const blkptr_t *bp); typedef void (*spa_remap_cb_t)(uint64_t vdev, uint64_t offset, uint64_t size, void *arg); extern boolean_t spa_remap_blkptr(spa_t *spa, blkptr_t *bp, spa_remap_cb_t callback, void *arg); extern uint64_t spa_get_last_removal_txg(spa_t *spa); extern boolean_t spa_trust_config(spa_t *spa); extern uint64_t spa_missing_tvds_allowed(spa_t *spa); extern void spa_set_missing_tvds(spa_t *spa, uint64_t missing); extern boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa); extern uint64_t spa_total_metaslabs(spa_t *spa); extern boolean_t spa_multihost(spa_t *spa); extern uint32_t spa_get_hostid(spa_t *spa); extern void spa_activate_allocation_classes(spa_t *, dmu_tx_t *); extern boolean_t spa_livelist_delete_check(spa_t *spa); extern spa_mode_t spa_mode(spa_t *spa); extern uint64_t zfs_strtonum(const char *str, char **nptr); extern char *spa_his_ievent_table[]; extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, char *his_buf); extern int spa_history_log(spa_t *spa, const char *his_buf); extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl); extern void spa_history_log_version(spa_t *spa, const char *operation, dmu_tx_t *tx); extern void spa_history_log_internal(spa_t *spa, const char *operation, dmu_tx_t *tx, const char *fmt, ...) __printflike(4, 5); extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op, dmu_tx_t *tx, const char *fmt, ...) __printflike(4, 5); extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, dmu_tx_t *tx, const char *fmt, ...) __printflike(4, 5); extern const char *spa_state_to_name(spa_t *spa); /* error handling */ struct zbookmark_phys; extern void spa_log_error(spa_t *spa, const zbookmark_phys_t *zb); extern void spa_remove_error(spa_t *spa, zbookmark_phys_t *zb); extern int zfs_ereport_post(const char *clazz, spa_t *spa, vdev_t *vd, const zbookmark_phys_t *zb, zio_t *zio, uint64_t state); extern boolean_t zfs_ereport_is_valid(const char *clazz, spa_t *spa, vdev_t *vd, zio_t *zio); extern void zfs_ereport_taskq_fini(void); extern void zfs_ereport_clear(spa_t *spa, vdev_t *vd); extern nvlist_t *zfs_event_create(spa_t *spa, vdev_t *vd, const char *type, const char *name, nvlist_t *aux); extern void zfs_post_remove(spa_t *spa, vdev_t *vd); extern void zfs_post_state_change(spa_t *spa, vdev_t *vd, uint64_t laststate); extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd); extern uint64_t spa_get_errlog_size(spa_t *spa); extern int spa_get_errlog(spa_t *spa, void *uaddr, uint64_t *count); extern void spa_errlog_rotate(spa_t *spa); extern void spa_errlog_drain(spa_t *spa); extern void spa_errlog_sync(spa_t *spa, uint64_t txg); extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub); extern void spa_delete_dataset_errlog(spa_t *spa, uint64_t ds, dmu_tx_t *tx); extern void spa_swap_errlog(spa_t *spa, uint64_t new_head_ds, uint64_t old_head_ds, dmu_tx_t *tx); extern void sync_error_list(spa_t *spa, avl_tree_t *t, uint64_t *obj, dmu_tx_t *tx); extern void spa_upgrade_errlog(spa_t *spa, dmu_tx_t *tx); /* vdev cache */ extern void vdev_cache_stat_init(void); extern void vdev_cache_stat_fini(void); /* vdev mirror */ extern void vdev_mirror_stat_init(void); extern void vdev_mirror_stat_fini(void); /* Initialization and termination */ extern void spa_init(spa_mode_t mode); extern void spa_fini(void); extern void spa_boot_init(void); /* properties */ extern int spa_prop_set(spa_t *spa, nvlist_t *nvp); extern int spa_prop_get(spa_t *spa, nvlist_t **nvp); extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx); extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t); /* asynchronous event notification */ extern void spa_event_notify(spa_t *spa, vdev_t *vdev, nvlist_t *hist_nvl, const char *name); extern void zfs_ereport_zvol_post(const char *subclass, const char *name, const char *device_name, const char *raw_name); /* waiting for pool activities to complete */ extern int spa_wait(const char *pool, zpool_wait_activity_t activity, boolean_t *waited); extern int spa_wait_tag(const char *name, zpool_wait_activity_t activity, uint64_t tag, boolean_t *waited); extern void spa_notify_waiters(spa_t *spa); extern void spa_wake_waiters(spa_t *spa); extern void spa_import_os(spa_t *spa); extern void spa_export_os(spa_t *spa); extern void spa_activate_os(spa_t *spa); extern void spa_deactivate_os(spa_t *spa); /* module param call functions */ int param_set_deadman_ziotime(ZFS_MODULE_PARAM_ARGS); int param_set_deadman_synctime(ZFS_MODULE_PARAM_ARGS); int param_set_slop_shift(ZFS_MODULE_PARAM_ARGS); int param_set_deadman_failmode(ZFS_MODULE_PARAM_ARGS); #ifdef ZFS_DEBUG #define dprintf_bp(bp, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \ snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \ dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \ kmem_free(__blkbuf, BP_SPRINTF_LEN); \ } \ } while (0) #else #define dprintf_bp(bp, fmt, ...) #endif extern spa_mode_t spa_mode_global; extern int zfs_deadman_enabled; extern unsigned long zfs_deadman_synctime_ms; extern unsigned long zfs_deadman_ziotime_ms; extern unsigned long zfs_deadman_checktime_ms; extern kmem_cache_t *zio_buf_cache[]; extern kmem_cache_t *zio_data_buf_cache[]; #ifdef __cplusplus } #endif #endif /* _SYS_SPA_H */ diff --git a/include/sys/vdev.h b/include/sys/vdev.h index 5fec1d51a5f2..7a7c70dc1598 100644 --- a/include/sys/vdev.h +++ b/include/sys/vdev.h @@ -1,229 +1,232 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Datto Inc. All rights reserved. */ #ifndef _SYS_VDEV_H #define _SYS_VDEV_H #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif typedef enum vdev_dtl_type { DTL_MISSING, /* 0% replication: no copies of the data */ DTL_PARTIAL, /* less than 100% replication: some copies missing */ DTL_SCRUB, /* unable to fully repair during scrub/resilver */ DTL_OUTAGE, /* temporarily missing (used to attempt detach) */ DTL_TYPES } vdev_dtl_type_t; extern int zfs_nocacheflush; typedef boolean_t vdev_open_children_func_t(vdev_t *vd); extern void vdev_dbgmsg(vdev_t *vd, const char *fmt, ...) __attribute__((format(printf, 2, 3))); extern void vdev_dbgmsg_print_tree(vdev_t *, int); extern int vdev_open(vdev_t *); extern void vdev_open_children(vdev_t *); extern void vdev_open_children_subset(vdev_t *, vdev_open_children_func_t *); extern int vdev_validate(vdev_t *); extern int vdev_copy_path_strict(vdev_t *, vdev_t *); extern void vdev_copy_path_relaxed(vdev_t *, vdev_t *); extern void vdev_close(vdev_t *); extern int vdev_create(vdev_t *, uint64_t txg, boolean_t isreplace); extern void vdev_reopen(vdev_t *); extern int vdev_validate_aux(vdev_t *vd); extern zio_t *vdev_probe(vdev_t *vd, zio_t *pio); extern boolean_t vdev_is_concrete(vdev_t *vd); extern boolean_t vdev_is_bootable(vdev_t *vd); extern vdev_t *vdev_lookup_top(spa_t *spa, uint64_t vdev); extern vdev_t *vdev_lookup_by_guid(vdev_t *vd, uint64_t guid); extern int vdev_count_leaves(spa_t *spa); extern void vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t d, uint64_t txg, uint64_t size); extern boolean_t vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t d, uint64_t txg, uint64_t size); extern boolean_t vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t d); extern boolean_t vdev_default_need_resilver(vdev_t *vd, const dva_t *dva, size_t psize, uint64_t phys_birth); extern boolean_t vdev_dtl_need_resilver(vdev_t *vd, const dva_t *dva, size_t psize, uint64_t phys_birth); extern void vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, boolean_t scrub_done, boolean_t rebuild_done); extern boolean_t vdev_dtl_required(vdev_t *vd); extern boolean_t vdev_resilver_needed(vdev_t *vd, uint64_t *minp, uint64_t *maxp); extern void vdev_destroy_unlink_zap(vdev_t *vd, uint64_t zapobj, dmu_tx_t *tx); extern uint64_t vdev_create_link_zap(vdev_t *vd, dmu_tx_t *tx); extern void vdev_construct_zaps(vdev_t *vd, dmu_tx_t *tx); extern void vdev_destroy_spacemaps(vdev_t *vd, dmu_tx_t *tx); extern void vdev_indirect_mark_obsolete(vdev_t *vd, uint64_t offset, uint64_t size); extern void spa_vdev_indirect_mark_obsolete(spa_t *spa, uint64_t vdev, uint64_t offset, uint64_t size, dmu_tx_t *tx); extern boolean_t vdev_replace_in_progress(vdev_t *vdev); extern void vdev_hold(vdev_t *); extern void vdev_rele(vdev_t *); extern int vdev_metaslab_init(vdev_t *vd, uint64_t txg); extern void vdev_metaslab_fini(vdev_t *vd); extern void vdev_metaslab_set_size(vdev_t *); extern void vdev_expand(vdev_t *vd, uint64_t txg); extern void vdev_split(vdev_t *vd); extern void vdev_deadman(vdev_t *vd, const char *tag); typedef void vdev_xlate_func_t(void *arg, range_seg64_t *physical_rs); extern boolean_t vdev_xlate_is_empty(range_seg64_t *rs); extern void vdev_xlate(vdev_t *vd, const range_seg64_t *logical_rs, range_seg64_t *physical_rs, range_seg64_t *remain_rs); extern void vdev_xlate_walk(vdev_t *vd, const range_seg64_t *logical_rs, vdev_xlate_func_t *func, void *arg); extern void vdev_get_stats_ex(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx); extern metaslab_group_t *vdev_get_mg(vdev_t *vd, metaslab_class_t *mc); extern void vdev_get_stats(vdev_t *vd, vdev_stat_t *vs); extern void vdev_clear_stats(vdev_t *vd); extern void vdev_stat_update(zio_t *zio, uint64_t psize); extern void vdev_scan_stat_init(vdev_t *vd); extern void vdev_propagate_state(vdev_t *vd); extern void vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state, vdev_aux_t aux); extern boolean_t vdev_children_are_offline(vdev_t *vd); extern void vdev_space_update(vdev_t *vd, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta); extern int64_t vdev_deflated_space(vdev_t *vd, int64_t space); extern uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize); /* * Return the amount of space allocated for a gang block header. */ static inline uint64_t vdev_gang_header_asize(vdev_t *vd) { return (vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE)); } extern int vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux); extern int vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux); extern int vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *); extern int vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags); +extern int vdev_remove_wanted(spa_t *spa, uint64_t guid); extern void vdev_clear(spa_t *spa, vdev_t *vd); extern boolean_t vdev_is_dead(vdev_t *vd); extern boolean_t vdev_readable(vdev_t *vd); extern boolean_t vdev_writeable(vdev_t *vd); extern boolean_t vdev_allocatable(vdev_t *vd); extern boolean_t vdev_accessible(vdev_t *vd, zio_t *zio); extern boolean_t vdev_is_spacemap_addressable(vdev_t *vd); extern void vdev_cache_init(vdev_t *vd); extern void vdev_cache_fini(vdev_t *vd); extern boolean_t vdev_cache_read(zio_t *zio); extern void vdev_cache_write(zio_t *zio); extern void vdev_cache_purge(vdev_t *vd); extern void vdev_queue_init(vdev_t *vd); extern void vdev_queue_fini(vdev_t *vd); extern zio_t *vdev_queue_io(zio_t *zio); extern void vdev_queue_io_done(zio_t *zio); extern void vdev_queue_change_io_priority(zio_t *zio, zio_priority_t priority); extern int vdev_queue_length(vdev_t *vd); extern uint64_t vdev_queue_last_offset(vdev_t *vd); extern void vdev_config_dirty(vdev_t *vd); extern void vdev_config_clean(vdev_t *vd); extern int vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg); extern void vdev_state_dirty(vdev_t *vd); extern void vdev_state_clean(vdev_t *vd); extern void vdev_defer_resilver(vdev_t *vd); extern boolean_t vdev_clear_resilver_deferred(vdev_t *vd, dmu_tx_t *tx); typedef enum vdev_config_flag { VDEV_CONFIG_SPARE = 1 << 0, VDEV_CONFIG_L2CACHE = 1 << 1, VDEV_CONFIG_REMOVING = 1 << 2, VDEV_CONFIG_MOS = 1 << 3, VDEV_CONFIG_MISSING = 1 << 4 } vdev_config_flag_t; +extern void vdev_post_kobj_evt(vdev_t *vd); +extern void vdev_clear_kobj_evt(vdev_t *vd); extern void vdev_top_config_generate(spa_t *spa, nvlist_t *config); extern nvlist_t *vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats, vdev_config_flag_t flags); /* * Label routines */ struct uberblock; extern uint64_t vdev_label_offset(uint64_t psize, int l, uint64_t offset); extern int vdev_label_number(uint64_t psise, uint64_t offset); extern nvlist_t *vdev_label_read_config(vdev_t *vd, uint64_t txg); extern void vdev_uberblock_load(vdev_t *, struct uberblock *, nvlist_t **); extern void vdev_config_generate_stats(vdev_t *vd, nvlist_t *nv); extern void vdev_label_write(zio_t *zio, vdev_t *vd, int l, abd_t *buf, uint64_t offset, uint64_t size, zio_done_func_t *done, void *priv, int flags); extern int vdev_label_read_bootenv(vdev_t *, nvlist_t *); extern int vdev_label_write_bootenv(vdev_t *, nvlist_t *); typedef enum { VDEV_LABEL_CREATE, /* create/add a new device */ VDEV_LABEL_REPLACE, /* replace an existing device */ VDEV_LABEL_SPARE, /* add a new hot spare */ VDEV_LABEL_REMOVE, /* remove an existing device */ VDEV_LABEL_L2CACHE, /* add an L2ARC cache device */ VDEV_LABEL_SPLIT /* generating new label for split-off dev */ } vdev_labeltype_t; extern int vdev_label_init(vdev_t *vd, uint64_t txg, vdev_labeltype_t reason); extern int vdev_prop_set(vdev_t *vd, nvlist_t *innvl, nvlist_t *outnvl); extern int vdev_prop_get(vdev_t *vd, nvlist_t *nvprops, nvlist_t *outnvl); #ifdef __cplusplus } #endif #endif /* _SYS_VDEV_H */ diff --git a/include/sys/vdev_impl.h b/include/sys/vdev_impl.h index 84c7363ddef0..b789d2c05d59 100644 --- a/include/sys/vdev_impl.h +++ b/include/sys/vdev_impl.h @@ -1,658 +1,661 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2017, Intel Corporation. */ #ifndef _SYS_VDEV_IMPL_H #define _SYS_VDEV_IMPL_H #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Virtual device descriptors. * * All storage pool operations go through the virtual device framework, * which provides data replication and I/O scheduling. */ /* * Forward declarations that lots of things need. */ typedef struct vdev_queue vdev_queue_t; typedef struct vdev_cache vdev_cache_t; typedef struct vdev_cache_entry vdev_cache_entry_t; struct abd; extern uint_t zfs_vdev_queue_depth_pct; extern uint_t zfs_vdev_def_queue_depth; extern uint_t zfs_vdev_async_write_max_active; /* * Virtual device operations */ typedef int vdev_init_func_t(spa_t *spa, nvlist_t *nv, void **tsd); +typedef void vdev_kobj_post_evt_func_t(vdev_t *vd); typedef void vdev_fini_func_t(vdev_t *vd); typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size, uint64_t *ashift, uint64_t *pshift); typedef void vdev_close_func_t(vdev_t *vd); typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize); typedef uint64_t vdev_min_asize_func_t(vdev_t *vd); typedef uint64_t vdev_min_alloc_func_t(vdev_t *vd); typedef void vdev_io_start_func_t(zio_t *zio); typedef void vdev_io_done_func_t(zio_t *zio); typedef void vdev_state_change_func_t(vdev_t *vd, int, int); typedef boolean_t vdev_need_resilver_func_t(vdev_t *vd, const dva_t *dva, size_t psize, uint64_t phys_birth); typedef void vdev_hold_func_t(vdev_t *vd); typedef void vdev_rele_func_t(vdev_t *vd); typedef void vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd, uint64_t offset, uint64_t size, void *arg); typedef void vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size, vdev_remap_cb_t callback, void *arg); /* * Given a target vdev, translates the logical range "in" to the physical * range "res" */ typedef void vdev_xlation_func_t(vdev_t *cvd, const range_seg64_t *logical, range_seg64_t *physical, range_seg64_t *remain); typedef uint64_t vdev_rebuild_asize_func_t(vdev_t *vd, uint64_t start, uint64_t size, uint64_t max_segment); typedef void vdev_metaslab_init_func_t(vdev_t *vd, uint64_t *startp, uint64_t *sizep); typedef void vdev_config_generate_func_t(vdev_t *vd, nvlist_t *nv); typedef uint64_t vdev_nparity_func_t(vdev_t *vd); typedef uint64_t vdev_ndisks_func_t(vdev_t *vd); typedef const struct vdev_ops { vdev_init_func_t *vdev_op_init; vdev_fini_func_t *vdev_op_fini; vdev_open_func_t *vdev_op_open; vdev_close_func_t *vdev_op_close; vdev_asize_func_t *vdev_op_asize; vdev_min_asize_func_t *vdev_op_min_asize; vdev_min_alloc_func_t *vdev_op_min_alloc; vdev_io_start_func_t *vdev_op_io_start; vdev_io_done_func_t *vdev_op_io_done; vdev_state_change_func_t *vdev_op_state_change; vdev_need_resilver_func_t *vdev_op_need_resilver; vdev_hold_func_t *vdev_op_hold; vdev_rele_func_t *vdev_op_rele; vdev_remap_func_t *vdev_op_remap; vdev_xlation_func_t *vdev_op_xlate; vdev_rebuild_asize_func_t *vdev_op_rebuild_asize; vdev_metaslab_init_func_t *vdev_op_metaslab_init; vdev_config_generate_func_t *vdev_op_config_generate; vdev_nparity_func_t *vdev_op_nparity; vdev_ndisks_func_t *vdev_op_ndisks; + vdev_kobj_post_evt_func_t *vdev_op_kobj_evt_post; char vdev_op_type[16]; boolean_t vdev_op_leaf; } vdev_ops_t; /* * Virtual device properties */ struct vdev_cache_entry { struct abd *ve_abd; uint64_t ve_offset; clock_t ve_lastused; avl_node_t ve_offset_node; avl_node_t ve_lastused_node; uint32_t ve_hits; uint16_t ve_missed_update; zio_t *ve_fill_io; }; struct vdev_cache { avl_tree_t vc_offset_tree; avl_tree_t vc_lastused_tree; kmutex_t vc_lock; }; typedef struct vdev_queue_class { uint32_t vqc_active; /* * Sorted by offset or timestamp, depending on if the queue is * LBA-ordered vs FIFO. */ avl_tree_t vqc_queued_tree; } vdev_queue_class_t; struct vdev_queue { vdev_t *vq_vdev; vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE]; avl_tree_t vq_active_tree; avl_tree_t vq_read_offset_tree; avl_tree_t vq_write_offset_tree; avl_tree_t vq_trim_offset_tree; uint64_t vq_last_offset; zio_priority_t vq_last_prio; /* Last sent I/O priority. */ uint32_t vq_ia_active; /* Active interactive I/Os. */ uint32_t vq_nia_credit; /* Non-interactive I/Os credit. */ hrtime_t vq_io_complete_ts; /* time last i/o completed */ hrtime_t vq_io_delta_ts; zio_t vq_io_search; /* used as local for stack reduction */ kmutex_t vq_lock; }; typedef enum vdev_alloc_bias { VDEV_BIAS_NONE, VDEV_BIAS_LOG, /* dedicated to ZIL data (SLOG) */ VDEV_BIAS_SPECIAL, /* dedicated to ddt, metadata, and small blks */ VDEV_BIAS_DEDUP /* dedicated to dedup metadata */ } vdev_alloc_bias_t; /* * On-disk indirect vdev state. * * An indirect vdev is described exclusively in the MOS config of a pool. * The config for an indirect vdev includes several fields, which are * accessed in memory by a vdev_indirect_config_t. */ typedef struct vdev_indirect_config { /* * Object (in MOS) which contains the indirect mapping. This object * contains an array of vdev_indirect_mapping_entry_phys_t ordered by * vimep_src. The bonus buffer for this object is a * vdev_indirect_mapping_phys_t. This object is allocated when a vdev * removal is initiated. * * Note that this object can be empty if none of the data on the vdev * has been copied yet. */ uint64_t vic_mapping_object; /* * Object (in MOS) which contains the birth times for the mapping * entries. This object contains an array of * vdev_indirect_birth_entry_phys_t sorted by vibe_offset. The bonus * buffer for this object is a vdev_indirect_birth_phys_t. This object * is allocated when a vdev removal is initiated. * * Note that this object can be empty if none of the vdev has yet been * copied. */ uint64_t vic_births_object; /* * This is the vdev ID which was removed previous to this vdev, or * UINT64_MAX if there are no previously removed vdevs. */ uint64_t vic_prev_indirect_vdev; } vdev_indirect_config_t; /* * Virtual device descriptor */ struct vdev { /* * Common to all vdev types. */ uint64_t vdev_id; /* child number in vdev parent */ uint64_t vdev_guid; /* unique ID for this vdev */ uint64_t vdev_guid_sum; /* self guid + all child guids */ uint64_t vdev_orig_guid; /* orig. guid prior to remove */ uint64_t vdev_asize; /* allocatable device capacity */ uint64_t vdev_min_asize; /* min acceptable asize */ uint64_t vdev_max_asize; /* max acceptable asize */ uint64_t vdev_ashift; /* block alignment shift */ /* * Logical block alignment shift * * The smallest sized/aligned I/O supported by the device. */ uint64_t vdev_logical_ashift; /* * Physical block alignment shift * * The device supports logical I/Os with vdev_logical_ashift * size/alignment, but optimum performance will be achieved by * aligning/sizing requests to vdev_physical_ashift. Smaller * requests may be inflated or incur device level read-modify-write * operations. * * May be 0 to indicate no preference (i.e. use vdev_logical_ashift). */ uint64_t vdev_physical_ashift; uint64_t vdev_state; /* see VDEV_STATE_* #defines */ uint64_t vdev_prevstate; /* used when reopening a vdev */ vdev_ops_t *vdev_ops; /* vdev operations */ spa_t *vdev_spa; /* spa for this vdev */ void *vdev_tsd; /* type-specific data */ vdev_t *vdev_top; /* top-level vdev */ vdev_t *vdev_parent; /* parent vdev */ vdev_t **vdev_child; /* array of children */ uint64_t vdev_children; /* number of children */ vdev_stat_t vdev_stat; /* virtual device statistics */ vdev_stat_ex_t vdev_stat_ex; /* extended statistics */ boolean_t vdev_expanding; /* expand the vdev? */ boolean_t vdev_reopening; /* reopen in progress? */ boolean_t vdev_nonrot; /* true if solid state */ int vdev_load_error; /* error on last load */ int vdev_open_error; /* error on last open */ int vdev_validate_error; /* error on last validate */ kthread_t *vdev_open_thread; /* thread opening children */ kthread_t *vdev_validate_thread; /* thread validating children */ uint64_t vdev_crtxg; /* txg when top-level was added */ /* * Top-level vdev state. */ uint64_t vdev_ms_array; /* metaslab array object */ uint64_t vdev_ms_shift; /* metaslab size shift */ uint64_t vdev_ms_count; /* number of metaslabs */ metaslab_group_t *vdev_mg; /* metaslab group */ metaslab_group_t *vdev_log_mg; /* embedded slog metaslab group */ metaslab_t **vdev_ms; /* metaslab array */ uint64_t vdev_pending_fastwrite; /* allocated fastwrites */ txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */ txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */ txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */ boolean_t vdev_remove_wanted; /* async remove wanted? */ boolean_t vdev_probe_wanted; /* async probe wanted? */ list_node_t vdev_config_dirty_node; /* config dirty list */ list_node_t vdev_state_dirty_node; /* state dirty list */ uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */ uint64_t vdev_islog; /* is an intent log device */ uint64_t vdev_noalloc; /* device is passivated? */ uint64_t vdev_removing; /* device is being removed? */ boolean_t vdev_ishole; /* is a hole in the namespace */ uint64_t vdev_top_zap; vdev_alloc_bias_t vdev_alloc_bias; /* metaslab allocation bias */ /* pool checkpoint related */ space_map_t *vdev_checkpoint_sm; /* contains reserved blocks */ /* Initialize related */ boolean_t vdev_initialize_exit_wanted; vdev_initializing_state_t vdev_initialize_state; list_node_t vdev_initialize_node; kthread_t *vdev_initialize_thread; /* Protects vdev_initialize_thread and vdev_initialize_state. */ kmutex_t vdev_initialize_lock; kcondvar_t vdev_initialize_cv; uint64_t vdev_initialize_offset[TXG_SIZE]; uint64_t vdev_initialize_last_offset; range_tree_t *vdev_initialize_tree; /* valid while initializing */ uint64_t vdev_initialize_bytes_est; uint64_t vdev_initialize_bytes_done; uint64_t vdev_initialize_action_time; /* start and end time */ /* TRIM related */ boolean_t vdev_trim_exit_wanted; boolean_t vdev_autotrim_exit_wanted; vdev_trim_state_t vdev_trim_state; list_node_t vdev_trim_node; kmutex_t vdev_autotrim_lock; kcondvar_t vdev_autotrim_cv; kthread_t *vdev_autotrim_thread; /* Protects vdev_trim_thread and vdev_trim_state. */ kmutex_t vdev_trim_lock; kcondvar_t vdev_trim_cv; kthread_t *vdev_trim_thread; uint64_t vdev_trim_offset[TXG_SIZE]; uint64_t vdev_trim_last_offset; uint64_t vdev_trim_bytes_est; uint64_t vdev_trim_bytes_done; uint64_t vdev_trim_rate; /* requested rate (bytes/sec) */ uint64_t vdev_trim_partial; /* requested partial TRIM */ uint64_t vdev_trim_secure; /* requested secure TRIM */ uint64_t vdev_trim_action_time; /* start and end time */ /* Rebuild related */ boolean_t vdev_rebuilding; boolean_t vdev_rebuild_exit_wanted; boolean_t vdev_rebuild_cancel_wanted; boolean_t vdev_rebuild_reset_wanted; kmutex_t vdev_rebuild_lock; kcondvar_t vdev_rebuild_cv; kthread_t *vdev_rebuild_thread; vdev_rebuild_t vdev_rebuild_config; /* For limiting outstanding I/Os (initialize, TRIM) */ kmutex_t vdev_initialize_io_lock; kcondvar_t vdev_initialize_io_cv; uint64_t vdev_initialize_inflight; kmutex_t vdev_trim_io_lock; kcondvar_t vdev_trim_io_cv; uint64_t vdev_trim_inflight[3]; /* * Values stored in the config for an indirect or removing vdev. */ vdev_indirect_config_t vdev_indirect_config; /* * The vdev_indirect_rwlock protects the vdev_indirect_mapping * pointer from changing on indirect vdevs (when it is condensed). * Note that removing (not yet indirect) vdevs have different * access patterns (the mapping is not accessed from open context, * e.g. from zio_read) and locking strategy (e.g. svr_lock). */ krwlock_t vdev_indirect_rwlock; vdev_indirect_mapping_t *vdev_indirect_mapping; vdev_indirect_births_t *vdev_indirect_births; /* * In memory data structures used to manage the obsolete sm, for * indirect or removing vdevs. * * The vdev_obsolete_segments is the in-core record of the segments * that are no longer referenced anywhere in the pool (due to * being freed or remapped and not referenced by any snapshots). * During a sync, segments are added to vdev_obsolete_segments * via vdev_indirect_mark_obsolete(); at the end of each sync * pass, this is appended to vdev_obsolete_sm via * vdev_indirect_sync_obsolete(). The vdev_obsolete_lock * protects against concurrent modifications of vdev_obsolete_segments * from multiple zio threads. */ kmutex_t vdev_obsolete_lock; range_tree_t *vdev_obsolete_segments; space_map_t *vdev_obsolete_sm; /* * Protects the vdev_scan_io_queue field itself as well as the * structure's contents (when present). */ kmutex_t vdev_scan_io_queue_lock; struct dsl_scan_io_queue *vdev_scan_io_queue; /* * Leaf vdev state. */ range_tree_t *vdev_dtl[DTL_TYPES]; /* dirty time logs */ space_map_t *vdev_dtl_sm; /* dirty time log space map */ txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */ uint64_t vdev_dtl_object; /* DTL object */ uint64_t vdev_psize; /* physical device capacity */ uint64_t vdev_wholedisk; /* true if this is a whole disk */ uint64_t vdev_offline; /* persistent offline state */ uint64_t vdev_faulted; /* persistent faulted state */ uint64_t vdev_degraded; /* persistent degraded state */ uint64_t vdev_removed; /* persistent removed state */ uint64_t vdev_resilver_txg; /* persistent resilvering state */ uint64_t vdev_rebuild_txg; /* persistent rebuilding state */ char *vdev_path; /* vdev path (if any) */ char *vdev_devid; /* vdev devid (if any) */ char *vdev_physpath; /* vdev device path (if any) */ char *vdev_enc_sysfs_path; /* enclosure sysfs path */ char *vdev_fru; /* physical FRU location */ uint64_t vdev_not_present; /* not present during import */ uint64_t vdev_unspare; /* unspare when resilvering done */ boolean_t vdev_nowritecache; /* true if flushwritecache failed */ boolean_t vdev_has_trim; /* TRIM is supported */ boolean_t vdev_has_securetrim; /* secure TRIM is supported */ boolean_t vdev_checkremove; /* temporary online test */ boolean_t vdev_forcefault; /* force online fault */ boolean_t vdev_splitting; /* split or repair in progress */ boolean_t vdev_delayed_close; /* delayed device close? */ boolean_t vdev_tmpoffline; /* device taken offline temporarily? */ boolean_t vdev_detached; /* device detached? */ boolean_t vdev_cant_read; /* vdev is failing all reads */ boolean_t vdev_cant_write; /* vdev is failing all writes */ boolean_t vdev_isspare; /* was a hot spare */ boolean_t vdev_isl2cache; /* was a l2cache device */ boolean_t vdev_copy_uberblocks; /* post expand copy uberblocks */ boolean_t vdev_resilver_deferred; /* resilver deferred */ + boolean_t vdev_kobj_flag; /* kobj event record */ vdev_queue_t vdev_queue; /* I/O deadline schedule queue */ vdev_cache_t vdev_cache; /* physical block cache */ spa_aux_vdev_t *vdev_aux; /* for l2cache and spares vdevs */ zio_t *vdev_probe_zio; /* root of current probe */ vdev_aux_t vdev_label_aux; /* on-disk aux state */ uint64_t vdev_leaf_zap; hrtime_t vdev_mmp_pending; /* 0 if write finished */ uint64_t vdev_mmp_kstat_id; /* to find kstat entry */ uint64_t vdev_expansion_time; /* vdev's last expansion time */ list_node_t vdev_leaf_node; /* leaf vdev list */ /* * For DTrace to work in userland (libzpool) context, these fields must * remain at the end of the structure. DTrace will use the kernel's * CTF definition for 'struct vdev', and since the size of a kmutex_t is * larger in userland, the offsets for the rest of the fields would be * incorrect. */ kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */ kmutex_t vdev_stat_lock; /* vdev_stat */ kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */ /* * We rate limit ZIO delay, deadman, and checksum events, since they * can flood ZED with tons of events when a drive is acting up. */ zfs_ratelimit_t vdev_delay_rl; zfs_ratelimit_t vdev_deadman_rl; zfs_ratelimit_t vdev_checksum_rl; }; #define VDEV_PAD_SIZE (8 << 10) /* 2 padding areas (vl_pad1 and vl_be) to skip */ #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2 #define VDEV_PHYS_SIZE (112 << 10) #define VDEV_UBERBLOCK_RING (128 << 10) /* * MMP blocks occupy the last MMP_BLOCKS_PER_LABEL slots in the uberblock * ring when MMP is enabled. */ #define MMP_BLOCKS_PER_LABEL 1 /* The largest uberblock we support is 8k. */ #define MAX_UBERBLOCK_SHIFT (13) #define VDEV_UBERBLOCK_SHIFT(vd) \ MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \ MAX_UBERBLOCK_SHIFT) #define VDEV_UBERBLOCK_COUNT(vd) \ (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd)) #define VDEV_UBERBLOCK_OFFSET(vd, n) \ offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)]) #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd)) typedef struct vdev_phys { char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)]; zio_eck_t vp_zbt; } vdev_phys_t; typedef enum vbe_vers { /* * The bootenv file is stored as ascii text in the envblock. * It is used by the GRUB bootloader used on Linux to store the * contents of the grubenv file. The file is stored as raw ASCII, * and is protected by an embedded checksum. By default, GRUB will * check if the boot filesystem supports storing the environment data * in a special location, and if so, will invoke filesystem specific * logic to retrieve it. This can be overridden by a variable, should * the user so desire. */ VB_RAW = 0, /* * The bootenv file is converted to an nvlist and then packed into the * envblock. */ VB_NVLIST = 1 } vbe_vers_t; typedef struct vdev_boot_envblock { uint64_t vbe_version; char vbe_bootenv[VDEV_PAD_SIZE - sizeof (uint64_t) - sizeof (zio_eck_t)]; zio_eck_t vbe_zbt; } vdev_boot_envblock_t; _Static_assert(sizeof (vdev_boot_envblock_t) == VDEV_PAD_SIZE, "vdev_boot_envblock_t wrong size"); typedef struct vdev_label { char vl_pad1[VDEV_PAD_SIZE]; /* 8K */ vdev_boot_envblock_t vl_be; /* 8K */ vdev_phys_t vl_vdev_phys; /* 112K */ char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */ } vdev_label_t; /* 256K total */ /* * vdev_dirty() flags */ #define VDD_METASLAB 0x01 #define VDD_DTL 0x02 /* Offset of embedded boot loader region on each label */ #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t)) /* * Size of embedded boot loader region on each label. * The total size of the first two labels plus the boot area is 4MB. */ #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */ /* * Size of label regions at the start and end of each leaf device. */ #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE) #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t)) #define VDEV_LABELS 4 #define VDEV_BEST_LABEL VDEV_LABELS #define VDEV_OFFSET_IS_LABEL(vd, off) \ (((off) < VDEV_LABEL_START_SIZE) || \ ((off) >= ((vd)->vdev_psize - VDEV_LABEL_END_SIZE))) #define VDEV_ALLOC_LOAD 0 #define VDEV_ALLOC_ADD 1 #define VDEV_ALLOC_SPARE 2 #define VDEV_ALLOC_L2CACHE 3 #define VDEV_ALLOC_ROOTPOOL 4 #define VDEV_ALLOC_SPLIT 5 #define VDEV_ALLOC_ATTACH 6 /* * Allocate or free a vdev */ extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops); extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config, vdev_t *parent, uint_t id, int alloctype); extern void vdev_free(vdev_t *vd); /* * Add or remove children and parents */ extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd); extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd); extern void vdev_compact_children(vdev_t *pvd); extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops); extern void vdev_remove_parent(vdev_t *cvd); /* * vdev sync load and sync */ extern boolean_t vdev_log_state_valid(vdev_t *vd); extern int vdev_load(vdev_t *vd); extern int vdev_dtl_load(vdev_t *vd); extern void vdev_sync(vdev_t *vd, uint64_t txg); extern void vdev_sync_done(vdev_t *vd, uint64_t txg); extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg); extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg); /* * Available vdev types. */ extern vdev_ops_t vdev_root_ops; extern vdev_ops_t vdev_mirror_ops; extern vdev_ops_t vdev_replacing_ops; extern vdev_ops_t vdev_raidz_ops; extern vdev_ops_t vdev_draid_ops; extern vdev_ops_t vdev_draid_spare_ops; extern vdev_ops_t vdev_disk_ops; extern vdev_ops_t vdev_file_ops; extern vdev_ops_t vdev_missing_ops; extern vdev_ops_t vdev_hole_ops; extern vdev_ops_t vdev_spare_ops; extern vdev_ops_t vdev_indirect_ops; /* * Common size functions */ extern void vdev_default_xlate(vdev_t *vd, const range_seg64_t *logical_rs, range_seg64_t *physical_rs, range_seg64_t *remain_rs); extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize); extern uint64_t vdev_default_min_asize(vdev_t *vd); extern uint64_t vdev_get_min_asize(vdev_t *vd); extern void vdev_set_min_asize(vdev_t *vd); extern uint64_t vdev_get_min_alloc(vdev_t *vd); extern uint64_t vdev_get_nparity(vdev_t *vd); extern uint64_t vdev_get_ndisks(vdev_t *vd); /* * Global variables */ extern int zfs_vdev_standard_sm_blksz; /* * Functions from vdev_indirect.c */ extern void vdev_indirect_sync_obsolete(vdev_t *vd, dmu_tx_t *tx); extern boolean_t vdev_indirect_should_condense(vdev_t *vd); extern void spa_condense_indirect_start_sync(vdev_t *vd, dmu_tx_t *tx); extern int vdev_obsolete_sm_object(vdev_t *vd, uint64_t *sm_obj); extern int vdev_obsolete_counts_are_precise(vdev_t *vd, boolean_t *are_precise); /* * Other miscellaneous functions */ int vdev_checkpoint_sm_object(vdev_t *vd, uint64_t *sm_obj); void vdev_metaslab_group_create(vdev_t *vd); uint64_t vdev_best_ashift(uint64_t logical, uint64_t a, uint64_t b); /* * Vdev ashift optimization tunables */ extern uint64_t zfs_vdev_min_auto_ashift; extern uint64_t zfs_vdev_max_auto_ashift; int param_set_min_auto_ashift(ZFS_MODULE_PARAM_ARGS); int param_set_max_auto_ashift(ZFS_MODULE_PARAM_ARGS); #ifdef __cplusplus } #endif #endif /* _SYS_VDEV_IMPL_H */ diff --git a/include/sys/zfs_context.h b/include/sys/zfs_context.h index 83ed97fbec7f..1f7e8bffabdb 100644 --- a/include/sys/zfs_context.h +++ b/include/sys/zfs_context.h @@ -1,778 +1,779 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2018 by Delphix. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. */ #ifndef _SYS_ZFS_CONTEXT_H #define _SYS_ZFS_CONTEXT_H #ifdef __cplusplus extern "C" { #endif /* * This code compiles in three different contexts. When __KERNEL__ is defined, * the code uses "unix-like" kernel interfaces. When _STANDALONE is defined, the * code is running in a reduced capacity environment of the boot loader which is * generally a subset of both POSIX and kernel interfaces (with a few unique * interfaces too). When neither are defined, it's in a userland POSIX or * similar environment. */ #if defined(__KERNEL__) || defined(_STANDALONE) #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #else /* _KERNEL || _STANDALONE */ #define _SYS_MUTEX_H #define _SYS_RWLOCK_H #define _SYS_CONDVAR_H #define _SYS_VNODE_H #define _SYS_VFS_H #define _SYS_SUNDDI_H #define _SYS_CALLB_H #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Stack */ #define noinline __attribute__((noinline)) #define likely(x) __builtin_expect((x), 1) #define unlikely(x) __builtin_expect((x), 0) /* * Debugging */ /* * Note that we are not using the debugging levels. */ #define CE_CONT 0 /* continuation */ #define CE_NOTE 1 /* notice */ #define CE_WARN 2 /* warning */ #define CE_PANIC 3 /* panic */ #define CE_IGNORE 4 /* print nothing */ /* * ZFS debugging */ extern void dprintf_setup(int *argc, char **argv); extern void cmn_err(int, const char *, ...); extern void vcmn_err(int, const char *, va_list); extern __attribute__((noreturn)) void panic(const char *, ...); extern __attribute__((noreturn)) void vpanic(const char *, va_list); #define fm_panic panic /* * DTrace SDT probes have different signatures in userland than they do in * the kernel. If they're being used in kernel code, re-define them out of * existence for their counterparts in libzpool. * * Here's an example of how to use the set-error probes in userland: * zfs$target:::set-error /arg0 == EBUSY/ {stack();} * * Here's an example of how to use DTRACE_PROBE probes in userland: * If there is a probe declared as follows: * DTRACE_PROBE2(zfs__probe_name, uint64_t, blkid, dnode_t *, dn); * Then you can use it as follows: * zfs$target:::probe2 /copyinstr(arg0) == "zfs__probe_name"/ * {printf("%u %p\n", arg1, arg2);} */ #ifdef DTRACE_PROBE #undef DTRACE_PROBE #endif /* DTRACE_PROBE */ #define DTRACE_PROBE(a) #ifdef DTRACE_PROBE1 #undef DTRACE_PROBE1 #endif /* DTRACE_PROBE1 */ #define DTRACE_PROBE1(a, b, c) #ifdef DTRACE_PROBE2 #undef DTRACE_PROBE2 #endif /* DTRACE_PROBE2 */ #define DTRACE_PROBE2(a, b, c, d, e) #ifdef DTRACE_PROBE3 #undef DTRACE_PROBE3 #endif /* DTRACE_PROBE3 */ #define DTRACE_PROBE3(a, b, c, d, e, f, g) #ifdef DTRACE_PROBE4 #undef DTRACE_PROBE4 #endif /* DTRACE_PROBE4 */ #define DTRACE_PROBE4(a, b, c, d, e, f, g, h, i) /* * Tunables. */ typedef struct zfs_kernel_param { const char *name; /* unused stub */ } zfs_kernel_param_t; #define ZFS_MODULE_PARAM(scope_prefix, name_prefix, name, type, perm, desc) #define ZFS_MODULE_PARAM_ARGS void #define ZFS_MODULE_PARAM_CALL(scope_prefix, name_prefix, name, setfunc, \ getfunc, perm, desc) /* * Threads. */ typedef pthread_t kthread_t; #define TS_RUN 0x00000002 #define TS_JOINABLE 0x00000004 #define curthread ((void *)(uintptr_t)pthread_self()) #define getcomm() "unknown" #define thread_create_named(name, stk, stksize, func, arg, len, \ pp, state, pri) \ zk_thread_create(func, arg, stksize, state) #define thread_create(stk, stksize, func, arg, len, pp, state, pri) \ zk_thread_create(func, arg, stksize, state) #define thread_exit() pthread_exit(NULL) #define thread_join(t) pthread_join((pthread_t)(t), NULL) #define newproc(f, a, cid, pri, ctp, pid) (ENOSYS) /* in libzpool, p0 exists only to have its address taken */ typedef struct proc { uintptr_t this_is_never_used_dont_dereference_it; } proc_t; extern struct proc p0; #define curproc (&p0) #define PS_NONE -1 extern kthread_t *zk_thread_create(void (*func)(void *), void *arg, size_t stksize, int state); #define issig(why) (FALSE) #define ISSIG(thr, why) (FALSE) #define KPREEMPT_SYNC (-1) #define kpreempt(x) sched_yield() #define kpreempt_disable() ((void)0) #define kpreempt_enable() ((void)0) /* * Mutexes */ typedef struct kmutex { pthread_mutex_t m_lock; pthread_t m_owner; } kmutex_t; #define MUTEX_DEFAULT 0 #define MUTEX_NOLOCKDEP MUTEX_DEFAULT #define MUTEX_HELD(mp) pthread_equal((mp)->m_owner, pthread_self()) #define MUTEX_NOT_HELD(mp) !MUTEX_HELD(mp) extern void mutex_init(kmutex_t *mp, char *name, int type, void *cookie); extern void mutex_destroy(kmutex_t *mp); extern void mutex_enter(kmutex_t *mp); extern void mutex_exit(kmutex_t *mp); extern int mutex_tryenter(kmutex_t *mp); #define NESTED_SINGLE 1 #define mutex_enter_nested(mp, class) mutex_enter(mp) /* * RW locks */ typedef struct krwlock { pthread_rwlock_t rw_lock; pthread_t rw_owner; uint_t rw_readers; } krwlock_t; typedef int krw_t; #define RW_READER 0 #define RW_WRITER 1 #define RW_DEFAULT RW_READER #define RW_NOLOCKDEP RW_READER #define RW_READ_HELD(rw) ((rw)->rw_readers > 0) #define RW_WRITE_HELD(rw) pthread_equal((rw)->rw_owner, pthread_self()) #define RW_LOCK_HELD(rw) (RW_READ_HELD(rw) || RW_WRITE_HELD(rw)) extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg); extern void rw_destroy(krwlock_t *rwlp); extern void rw_enter(krwlock_t *rwlp, krw_t rw); extern int rw_tryenter(krwlock_t *rwlp, krw_t rw); extern int rw_tryupgrade(krwlock_t *rwlp); extern void rw_exit(krwlock_t *rwlp); #define rw_downgrade(rwlp) do { } while (0) /* * Credentials */ extern uid_t crgetuid(cred_t *cr); extern uid_t crgetruid(cred_t *cr); extern gid_t crgetgid(cred_t *cr); extern int crgetngroups(cred_t *cr); extern gid_t *crgetgroups(cred_t *cr); /* * Condition variables */ typedef pthread_cond_t kcondvar_t; #define CV_DEFAULT 0 #define CALLOUT_FLAG_ABSOLUTE 0x2 extern void cv_init(kcondvar_t *cv, char *name, int type, void *arg); extern void cv_destroy(kcondvar_t *cv); extern void cv_wait(kcondvar_t *cv, kmutex_t *mp); extern int cv_wait_sig(kcondvar_t *cv, kmutex_t *mp); extern int cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime); extern int cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res, int flag); extern void cv_signal(kcondvar_t *cv); extern void cv_broadcast(kcondvar_t *cv); #define cv_timedwait_io(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_timedwait_idle(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_timedwait_sig(cv, mp, at) cv_timedwait(cv, mp, at) #define cv_wait_io(cv, mp) cv_wait(cv, mp) #define cv_wait_idle(cv, mp) cv_wait(cv, mp) #define cv_wait_io_sig(cv, mp) cv_wait_sig(cv, mp) #define cv_timedwait_sig_hires(cv, mp, t, r, f) \ cv_timedwait_hires(cv, mp, t, r, f) #define cv_timedwait_idle_hires(cv, mp, t, r, f) \ cv_timedwait_hires(cv, mp, t, r, f) /* * Thread-specific data */ #define tsd_get(k) pthread_getspecific(k) #define tsd_set(k, v) pthread_setspecific(k, v) #define tsd_create(kp, d) pthread_key_create((pthread_key_t *)kp, d) #define tsd_destroy(kp) /* nothing */ #ifdef __FreeBSD__ typedef off_t loff_t; #endif /* * kstat creation, installation and deletion */ extern kstat_t *kstat_create(const char *, int, const char *, const char *, uchar_t, ulong_t, uchar_t); extern void kstat_install(kstat_t *); extern void kstat_delete(kstat_t *); extern void kstat_set_raw_ops(kstat_t *ksp, int (*headers)(char *buf, size_t size), int (*data)(char *buf, size_t size, void *data), void *(*addr)(kstat_t *ksp, loff_t index)); /* * procfs list manipulation */ typedef struct procfs_list { void *pl_private; kmutex_t pl_lock; list_t pl_list; uint64_t pl_next_id; size_t pl_node_offset; } procfs_list_t; #ifndef __cplusplus struct seq_file { }; void seq_printf(struct seq_file *m, const char *fmt, ...); typedef struct procfs_list_node { list_node_t pln_link; uint64_t pln_id; } procfs_list_node_t; void procfs_list_install(const char *module, const char *submodule, const char *name, mode_t mode, procfs_list_t *procfs_list, int (*show)(struct seq_file *f, void *p), int (*show_header)(struct seq_file *f), int (*clear)(procfs_list_t *procfs_list), size_t procfs_list_node_off); void procfs_list_uninstall(procfs_list_t *procfs_list); void procfs_list_destroy(procfs_list_t *procfs_list); void procfs_list_add(procfs_list_t *procfs_list, void *p); #endif /* * Kernel memory */ #define KM_SLEEP UMEM_NOFAIL #define KM_PUSHPAGE KM_SLEEP #define KM_NOSLEEP UMEM_DEFAULT #define KM_NORMALPRI 0 /* not needed with UMEM_DEFAULT */ #define KMC_NODEBUG UMC_NODEBUG #define KMC_KVMEM 0x0 #define kmem_alloc(_s, _f) umem_alloc(_s, _f) #define kmem_zalloc(_s, _f) umem_zalloc(_s, _f) #define kmem_free(_b, _s) umem_free(_b, _s) #define vmem_alloc(_s, _f) kmem_alloc(_s, _f) #define vmem_zalloc(_s, _f) kmem_zalloc(_s, _f) #define vmem_free(_b, _s) kmem_free(_b, _s) #define kmem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) \ umem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) #define kmem_cache_destroy(_c) umem_cache_destroy(_c) #define kmem_cache_alloc(_c, _f) umem_cache_alloc(_c, _f) #define kmem_cache_free(_c, _b) umem_cache_free(_c, _b) #define kmem_debugging() 0 #define kmem_cache_reap_now(_c) umem_cache_reap_now(_c); #define kmem_cache_set_move(_c, _cb) /* nothing */ #define POINTER_INVALIDATE(_pp) /* nothing */ #define POINTER_IS_VALID(_p) 0 typedef umem_cache_t kmem_cache_t; typedef enum kmem_cbrc { KMEM_CBRC_YES, KMEM_CBRC_NO, KMEM_CBRC_LATER, KMEM_CBRC_DONT_NEED, KMEM_CBRC_DONT_KNOW } kmem_cbrc_t; /* * Task queues */ #define TASKQ_NAMELEN 31 typedef uintptr_t taskqid_t; typedef void (task_func_t)(void *); typedef struct taskq_ent { struct taskq_ent *tqent_next; struct taskq_ent *tqent_prev; task_func_t *tqent_func; void *tqent_arg; uintptr_t tqent_flags; } taskq_ent_t; typedef struct taskq { char tq_name[TASKQ_NAMELEN + 1]; kmutex_t tq_lock; krwlock_t tq_threadlock; kcondvar_t tq_dispatch_cv; kcondvar_t tq_wait_cv; kthread_t **tq_threadlist; int tq_flags; int tq_active; int tq_nthreads; int tq_nalloc; int tq_minalloc; int tq_maxalloc; kcondvar_t tq_maxalloc_cv; int tq_maxalloc_wait; taskq_ent_t *tq_freelist; taskq_ent_t tq_task; } taskq_t; #define TQENT_FLAG_PREALLOC 0x1 /* taskq_dispatch_ent used */ #define TASKQ_PREPOPULATE 0x0001 #define TASKQ_CPR_SAFE 0x0002 /* Use CPR safe protocol */ #define TASKQ_DYNAMIC 0x0004 /* Use dynamic thread scheduling */ #define TASKQ_THREADS_CPU_PCT 0x0008 /* Scale # threads by # cpus */ #define TASKQ_DC_BATCH 0x0010 /* Mark threads as batch */ #define TQ_SLEEP KM_SLEEP /* Can block for memory */ #define TQ_NOSLEEP KM_NOSLEEP /* cannot block for memory; may fail */ #define TQ_NOQUEUE 0x02 /* Do not enqueue if can't dispatch */ #define TQ_FRONT 0x08 /* Queue in front */ #define TASKQID_INVALID ((taskqid_t)0) extern taskq_t *system_taskq; extern taskq_t *system_delay_taskq; extern taskq_t *taskq_create(const char *, int, pri_t, int, int, uint_t); #define taskq_create_proc(a, b, c, d, e, p, f) \ (taskq_create(a, b, c, d, e, f)) #define taskq_create_sysdc(a, b, d, e, p, dc, f) \ ((void) sizeof (dc), taskq_create(a, b, maxclsyspri, d, e, f)) extern taskqid_t taskq_dispatch(taskq_t *, task_func_t, void *, uint_t); extern taskqid_t taskq_dispatch_delay(taskq_t *, task_func_t, void *, uint_t, clock_t); extern void taskq_dispatch_ent(taskq_t *, task_func_t, void *, uint_t, taskq_ent_t *); extern int taskq_empty_ent(taskq_ent_t *); extern void taskq_init_ent(taskq_ent_t *); extern void taskq_destroy(taskq_t *); extern void taskq_wait(taskq_t *); extern void taskq_wait_id(taskq_t *, taskqid_t); extern void taskq_wait_outstanding(taskq_t *, taskqid_t); extern int taskq_member(taskq_t *, kthread_t *); extern taskq_t *taskq_of_curthread(void); extern int taskq_cancel_id(taskq_t *, taskqid_t); extern void system_taskq_init(void); extern void system_taskq_fini(void); #define XVA_MAPSIZE 3 #define XVA_MAGIC 0x78766174 extern char *vn_dumpdir; #define AV_SCANSTAMP_SZ 32 /* length of anti-virus scanstamp */ typedef struct xoptattr { inode_timespec_t xoa_createtime; /* Create time of file */ uint8_t xoa_archive; uint8_t xoa_system; uint8_t xoa_readonly; uint8_t xoa_hidden; uint8_t xoa_nounlink; uint8_t xoa_immutable; uint8_t xoa_appendonly; uint8_t xoa_nodump; uint8_t xoa_settable; uint8_t xoa_opaque; uint8_t xoa_av_quarantined; uint8_t xoa_av_modified; uint8_t xoa_av_scanstamp[AV_SCANSTAMP_SZ]; uint8_t xoa_reparse; uint8_t xoa_offline; uint8_t xoa_sparse; } xoptattr_t; typedef struct vattr { uint_t va_mask; /* bit-mask of attributes */ u_offset_t va_size; /* file size in bytes */ } vattr_t; typedef struct xvattr { vattr_t xva_vattr; /* Embedded vattr structure */ uint32_t xva_magic; /* Magic Number */ uint32_t xva_mapsize; /* Size of attr bitmap (32-bit words) */ uint32_t *xva_rtnattrmapp; /* Ptr to xva_rtnattrmap[] */ uint32_t xva_reqattrmap[XVA_MAPSIZE]; /* Requested attrs */ uint32_t xva_rtnattrmap[XVA_MAPSIZE]; /* Returned attrs */ xoptattr_t xva_xoptattrs; /* Optional attributes */ } xvattr_t; typedef struct vsecattr { uint_t vsa_mask; /* See below */ int vsa_aclcnt; /* ACL entry count */ void *vsa_aclentp; /* pointer to ACL entries */ int vsa_dfaclcnt; /* default ACL entry count */ void *vsa_dfaclentp; /* pointer to default ACL entries */ size_t vsa_aclentsz; /* ACE size in bytes of vsa_aclentp */ } vsecattr_t; #define AT_MODE 0x00002 #define AT_UID 0x00004 #define AT_GID 0x00008 #define AT_FSID 0x00010 #define AT_NODEID 0x00020 #define AT_NLINK 0x00040 #define AT_SIZE 0x00080 #define AT_ATIME 0x00100 #define AT_MTIME 0x00200 #define AT_CTIME 0x00400 #define AT_RDEV 0x00800 #define AT_BLKSIZE 0x01000 #define AT_NBLOCKS 0x02000 #define AT_SEQ 0x08000 #define AT_XVATTR 0x10000 #define CRCREAT 0 #define F_FREESP 11 #define FIGNORECASE 0x80000 /* request case-insensitive lookups */ /* * Random stuff */ #define ddi_get_lbolt() (gethrtime() >> 23) #define ddi_get_lbolt64() (gethrtime() >> 23) #define hz 119 /* frequency when using gethrtime() >> 23 for lbolt */ #define ddi_time_before(a, b) (a < b) #define ddi_time_after(a, b) ddi_time_before(b, a) #define ddi_time_before_eq(a, b) (!ddi_time_after(a, b)) #define ddi_time_after_eq(a, b) ddi_time_before_eq(b, a) #define ddi_time_before64(a, b) (a < b) #define ddi_time_after64(a, b) ddi_time_before64(b, a) #define ddi_time_before_eq64(a, b) (!ddi_time_after64(a, b)) #define ddi_time_after_eq64(a, b) ddi_time_before_eq64(b, a) extern void delay(clock_t ticks); #define SEC_TO_TICK(sec) ((sec) * hz) #define MSEC_TO_TICK(msec) (howmany((hrtime_t)(msec) * hz, MILLISEC)) #define USEC_TO_TICK(usec) (howmany((hrtime_t)(usec) * hz, MICROSEC)) #define NSEC_TO_TICK(nsec) (howmany((hrtime_t)(nsec) * hz, NANOSEC)) #define max_ncpus 64 #define boot_ncpus (sysconf(_SC_NPROCESSORS_ONLN)) /* * Process priorities as defined by setpriority(2) and getpriority(2). */ #define minclsyspri 19 #define maxclsyspri -20 #define defclsyspri 0 #define CPU_SEQID ((uintptr_t)pthread_self() & (max_ncpus - 1)) #define CPU_SEQID_UNSTABLE CPU_SEQID #define kcred NULL #define CRED() NULL #define ptob(x) ((x) * PAGESIZE) #define NN_DIVISOR_1000 (1U << 0) #define NN_NUMBUF_SZ (6) extern uint64_t physmem; extern const char *random_path; extern const char *urandom_path; extern int highbit64(uint64_t i); extern int lowbit64(uint64_t i); extern int random_get_bytes(uint8_t *ptr, size_t len); extern int random_get_pseudo_bytes(uint8_t *ptr, size_t len); static __inline__ uint32_t random_in_range(uint32_t range) { uint32_t r; ASSERT(range != 0); if (range == 1) return (0); (void) random_get_pseudo_bytes((uint8_t *)&r, sizeof (r)); return (r % range); } extern void kernel_init(int mode); extern void kernel_fini(void); extern void random_init(void); extern void random_fini(void); struct spa; extern void show_pool_stats(struct spa *); extern int set_global_var(char const *arg); typedef struct callb_cpr { kmutex_t *cc_lockp; } callb_cpr_t; #define CALLB_CPR_INIT(cp, lockp, func, name) { \ (cp)->cc_lockp = lockp; \ } #define CALLB_CPR_SAFE_BEGIN(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_SAFE_END(cp, lockp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ } #define CALLB_CPR_EXIT(cp) { \ ASSERT(MUTEX_HELD((cp)->cc_lockp)); \ mutex_exit((cp)->cc_lockp); \ } #define zone_dataset_visible(x, y) (1) #define INGLOBALZONE(z) (1) extern uint32_t zone_get_hostid(void *zonep); extern char *kmem_vasprintf(const char *fmt, va_list adx); extern char *kmem_asprintf(const char *fmt, ...); #define kmem_strfree(str) kmem_free((str), strlen(str) + 1) #define kmem_strdup(s) strdup(s) /* * Hostname information */ extern int ddi_strtoull(const char *str, char **nptr, int base, u_longlong_t *result); typedef struct utsname utsname_t; extern utsname_t *utsname(void); /* ZFS Boot Related stuff. */ struct _buf { intptr_t _fd; }; struct bootstat { uint64_t st_size; }; typedef struct ace_object { uid_t a_who; uint32_t a_access_mask; uint16_t a_flags; uint16_t a_type; uint8_t a_obj_type[16]; uint8_t a_inherit_obj_type[16]; } ace_object_t; #define ACE_ACCESS_ALLOWED_OBJECT_ACE_TYPE 0x05 #define ACE_ACCESS_DENIED_OBJECT_ACE_TYPE 0x06 #define ACE_SYSTEM_AUDIT_OBJECT_ACE_TYPE 0x07 #define ACE_SYSTEM_ALARM_OBJECT_ACE_TYPE 0x08 extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr); extern int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr); extern int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr); extern int secpolicy_zfs(const cred_t *cr); extern int secpolicy_zfs_proc(const cred_t *cr, proc_t *proc); extern zoneid_t getzoneid(void); /* SID stuff */ typedef struct ksiddomain { uint_t kd_ref; uint_t kd_len; char *kd_name; } ksiddomain_t; ksiddomain_t *ksid_lookupdomain(const char *); void ksiddomain_rele(ksiddomain_t *); #define DDI_SLEEP KM_SLEEP #define ddi_log_sysevent(_a, _b, _c, _d, _e, _f, _g) \ sysevent_post_event(_c, _d, _b, "libzpool", _e, _f) #define zfs_sleep_until(wakeup) \ do { \ hrtime_t delta = wakeup - gethrtime(); \ struct timespec ts; \ ts.tv_sec = delta / NANOSEC; \ ts.tv_nsec = delta % NANOSEC; \ (void) nanosleep(&ts, NULL); \ } while (0) typedef int fstrans_cookie_t; extern fstrans_cookie_t spl_fstrans_mark(void); extern void spl_fstrans_unmark(fstrans_cookie_t); extern int __spl_pf_fstrans_check(void); extern int kmem_cache_reap_active(void); /* * Kernel modules */ #define __init #define __exit #endif /* _KERNEL || _STANDALONE */ #ifdef __cplusplus }; #endif #endif /* _SYS_ZFS_CONTEXT_H */ diff --git a/lib/libzfs/libzfs.abi b/lib/libzfs/libzfs.abi index 7dd12df81718..3471bcac9412 100644 --- a/lib/libzfs/libzfs.abi +++ b/lib/libzfs/libzfs.abi @@ -1,5852 +1,5858 @@ + + + + + + diff --git a/lib/libzfs/libzfs_pool.c b/lib/libzfs/libzfs_pool.c index eea388cf348f..b9806dc30dac 100644 --- a/lib/libzfs/libzfs_pool.c +++ b/lib/libzfs/libzfs_pool.c @@ -1,5181 +1,5218 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2015 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright 2016 Igor Kozhukhov * Copyright (c) 2018 Datto Inc. * Copyright (c) 2017 Open-E, Inc. All Rights Reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2018, loli10K * Copyright (c) 2021, Colm Buckley * Copyright (c) 2021, Klara Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_namecheck.h" #include "zfs_prop.h" #include "libzfs_impl.h" #include "zfs_comutil.h" #include "zfeature_common.h" static boolean_t zpool_vdev_is_interior(const char *name); typedef struct prop_flags { int create:1; /* Validate property on creation */ int import:1; /* Validate property on import */ int vdevprop:1; /* Validate property as a VDEV property */ } prop_flags_t; /* * ==================================================================== * zpool property functions * ==================================================================== */ static int zpool_get_all_props(zpool_handle_t *zhp) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zcmd_alloc_dst_nvlist(hdl, &zc, 0); while (zfs_ioctl(hdl, ZFS_IOC_POOL_GET_PROPS, &zc) != 0) { if (errno == ENOMEM) zcmd_expand_dst_nvlist(hdl, &zc); else { zcmd_free_nvlists(&zc); return (-1); } } if (zcmd_read_dst_nvlist(hdl, &zc, &zhp->zpool_props) != 0) { zcmd_free_nvlists(&zc); return (-1); } zcmd_free_nvlists(&zc); return (0); } int zpool_props_refresh(zpool_handle_t *zhp) { nvlist_t *old_props; old_props = zhp->zpool_props; if (zpool_get_all_props(zhp) != 0) return (-1); nvlist_free(old_props); return (0); } static const char * zpool_get_prop_string(zpool_handle_t *zhp, zpool_prop_t prop, zprop_source_t *src) { nvlist_t *nv, *nvl; const char *value; zprop_source_t source; nvl = zhp->zpool_props; if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) { source = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); value = fnvlist_lookup_string(nv, ZPROP_VALUE); } else { source = ZPROP_SRC_DEFAULT; if ((value = zpool_prop_default_string(prop)) == NULL) value = "-"; } if (src) *src = source; return (value); } uint64_t zpool_get_prop_int(zpool_handle_t *zhp, zpool_prop_t prop, zprop_source_t *src) { nvlist_t *nv, *nvl; uint64_t value; zprop_source_t source; if (zhp->zpool_props == NULL && zpool_get_all_props(zhp)) { /* * zpool_get_all_props() has most likely failed because * the pool is faulted, but if all we need is the top level * vdev's guid then get it from the zhp config nvlist. */ if ((prop == ZPOOL_PROP_GUID) && (nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0) && (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value) == 0)) { return (value); } return (zpool_prop_default_numeric(prop)); } nvl = zhp->zpool_props; if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) { source = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); value = fnvlist_lookup_uint64(nv, ZPROP_VALUE); } else { source = ZPROP_SRC_DEFAULT; value = zpool_prop_default_numeric(prop); } if (src) *src = source; return (value); } /* * Map VDEV STATE to printed strings. */ const char * zpool_state_to_name(vdev_state_t state, vdev_aux_t aux) { switch (state) { case VDEV_STATE_CLOSED: case VDEV_STATE_OFFLINE: return (gettext("OFFLINE")); case VDEV_STATE_REMOVED: return (gettext("REMOVED")); case VDEV_STATE_CANT_OPEN: if (aux == VDEV_AUX_CORRUPT_DATA || aux == VDEV_AUX_BAD_LOG) return (gettext("FAULTED")); else if (aux == VDEV_AUX_SPLIT_POOL) return (gettext("SPLIT")); else return (gettext("UNAVAIL")); case VDEV_STATE_FAULTED: return (gettext("FAULTED")); case VDEV_STATE_DEGRADED: return (gettext("DEGRADED")); case VDEV_STATE_HEALTHY: return (gettext("ONLINE")); default: break; } return (gettext("UNKNOWN")); } /* * Map POOL STATE to printed strings. */ const char * zpool_pool_state_to_name(pool_state_t state) { switch (state) { default: break; case POOL_STATE_ACTIVE: return (gettext("ACTIVE")); case POOL_STATE_EXPORTED: return (gettext("EXPORTED")); case POOL_STATE_DESTROYED: return (gettext("DESTROYED")); case POOL_STATE_SPARE: return (gettext("SPARE")); case POOL_STATE_L2CACHE: return (gettext("L2CACHE")); case POOL_STATE_UNINITIALIZED: return (gettext("UNINITIALIZED")); case POOL_STATE_UNAVAIL: return (gettext("UNAVAIL")); case POOL_STATE_POTENTIALLY_ACTIVE: return (gettext("POTENTIALLY_ACTIVE")); } return (gettext("UNKNOWN")); } /* * Given a pool handle, return the pool health string ("ONLINE", "DEGRADED", * "SUSPENDED", etc). */ const char * zpool_get_state_str(zpool_handle_t *zhp) { zpool_errata_t errata; zpool_status_t status; const char *str; status = zpool_get_status(zhp, NULL, &errata); if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { str = gettext("FAULTED"); } else if (status == ZPOOL_STATUS_IO_FAILURE_WAIT || status == ZPOOL_STATUS_IO_FAILURE_MMP) { str = gettext("SUSPENDED"); } else { nvlist_t *nvroot = fnvlist_lookup_nvlist( zpool_get_config(zhp, NULL), ZPOOL_CONFIG_VDEV_TREE); uint_t vsc; vdev_stat_t *vs = (vdev_stat_t *)fnvlist_lookup_uint64_array( nvroot, ZPOOL_CONFIG_VDEV_STATS, &vsc); str = zpool_state_to_name(vs->vs_state, vs->vs_aux); } return (str); } /* * Get a zpool property value for 'prop' and return the value in * a pre-allocated buffer. */ int zpool_get_prop(zpool_handle_t *zhp, zpool_prop_t prop, char *buf, size_t len, zprop_source_t *srctype, boolean_t literal) { uint64_t intval; const char *strval; zprop_source_t src = ZPROP_SRC_NONE; if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { switch (prop) { case ZPOOL_PROP_NAME: (void) strlcpy(buf, zpool_get_name(zhp), len); break; case ZPOOL_PROP_HEALTH: (void) strlcpy(buf, zpool_get_state_str(zhp), len); break; case ZPOOL_PROP_GUID: intval = zpool_get_prop_int(zhp, prop, &src); (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); break; case ZPOOL_PROP_ALTROOT: case ZPOOL_PROP_CACHEFILE: case ZPOOL_PROP_COMMENT: case ZPOOL_PROP_COMPATIBILITY: if (zhp->zpool_props != NULL || zpool_get_all_props(zhp) == 0) { (void) strlcpy(buf, zpool_get_prop_string(zhp, prop, &src), len); break; } zfs_fallthrough; default: (void) strlcpy(buf, "-", len); break; } if (srctype != NULL) *srctype = src; return (0); } if (zhp->zpool_props == NULL && zpool_get_all_props(zhp) && prop != ZPOOL_PROP_NAME) return (-1); switch (zpool_prop_get_type(prop)) { case PROP_TYPE_STRING: (void) strlcpy(buf, zpool_get_prop_string(zhp, prop, &src), len); break; case PROP_TYPE_NUMBER: intval = zpool_get_prop_int(zhp, prop, &src); switch (prop) { case ZPOOL_PROP_SIZE: case ZPOOL_PROP_ALLOCATED: case ZPOOL_PROP_FREE: case ZPOOL_PROP_FREEING: case ZPOOL_PROP_LEAKED: case ZPOOL_PROP_ASHIFT: case ZPOOL_PROP_MAXBLOCKSIZE: case ZPOOL_PROP_MAXDNODESIZE: if (literal) (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); else (void) zfs_nicenum(intval, buf, len); break; case ZPOOL_PROP_EXPANDSZ: case ZPOOL_PROP_CHECKPOINT: if (intval == 0) { (void) strlcpy(buf, "-", len); } else if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) zfs_nicebytes(intval, buf, len); } break; case ZPOOL_PROP_CAPACITY: if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) snprintf(buf, len, "%llu%%", (u_longlong_t)intval); } break; case ZPOOL_PROP_FRAGMENTATION: if (intval == UINT64_MAX) { (void) strlcpy(buf, "-", len); } else if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) snprintf(buf, len, "%llu%%", (u_longlong_t)intval); } break; case ZPOOL_PROP_DEDUPRATIO: if (literal) (void) snprintf(buf, len, "%llu.%02llu", (u_longlong_t)(intval / 100), (u_longlong_t)(intval % 100)); else (void) snprintf(buf, len, "%llu.%02llux", (u_longlong_t)(intval / 100), (u_longlong_t)(intval % 100)); break; case ZPOOL_PROP_HEALTH: (void) strlcpy(buf, zpool_get_state_str(zhp), len); break; case ZPOOL_PROP_VERSION: if (intval >= SPA_VERSION_FEATURES) { (void) snprintf(buf, len, "-"); break; } zfs_fallthrough; default: (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } break; case PROP_TYPE_INDEX: intval = zpool_get_prop_int(zhp, prop, &src); if (zpool_prop_index_to_string(prop, intval, &strval) != 0) return (-1); (void) strlcpy(buf, strval, len); break; default: abort(); } if (srctype) *srctype = src; return (0); } /* * Check if the bootfs name has the same pool name as it is set to. * Assuming bootfs is a valid dataset name. */ static boolean_t bootfs_name_valid(const char *pool, const char *bootfs) { int len = strlen(pool); if (bootfs[0] == '\0') return (B_TRUE); if (!zfs_name_valid(bootfs, ZFS_TYPE_FILESYSTEM|ZFS_TYPE_SNAPSHOT)) return (B_FALSE); if (strncmp(pool, bootfs, len) == 0 && (bootfs[len] == '/' || bootfs[len] == '\0')) return (B_TRUE); return (B_FALSE); } /* * Given an nvlist of zpool properties to be set, validate that they are * correct, and parse any numeric properties (index, boolean, etc) if they are * specified as strings. */ static nvlist_t * zpool_valid_proplist(libzfs_handle_t *hdl, const char *poolname, nvlist_t *props, uint64_t version, prop_flags_t flags, char *errbuf) { nvpair_t *elem; nvlist_t *retprops; zpool_prop_t prop; char *strval; uint64_t intval; char *slash, *check; struct stat64 statbuf; zpool_handle_t *zhp; char report[1024]; if (nvlist_alloc(&retprops, NV_UNIQUE_NAME, 0) != 0) { (void) no_memory(hdl); return (NULL); } elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { const char *propname = nvpair_name(elem); if (flags.vdevprop && zpool_prop_vdev(propname)) { vdev_prop_t vprop = vdev_name_to_prop(propname); if (vdev_prop_readonly(vprop)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' " "is readonly"), propname); (void) zfs_error(hdl, EZFS_PROPREADONLY, errbuf); goto error; } if (zprop_parse_value(hdl, elem, vprop, ZFS_TYPE_VDEV, retprops, &strval, &intval, errbuf) != 0) goto error; continue; } else if (flags.vdevprop && vdev_prop_user(propname)) { if (nvlist_add_nvpair(retprops, elem) != 0) { (void) no_memory(hdl); goto error; } continue; } else if (flags.vdevprop) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid property: '%s'"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } prop = zpool_name_to_prop(propname); if (prop == ZPOOL_PROP_INVAL && zpool_prop_feature(propname)) { int err; char *fname = strchr(propname, '@') + 1; err = zfeature_lookup_name(fname, NULL); if (err != 0) { ASSERT3U(err, ==, ENOENT); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "feature '%s' unsupported by kernel"), fname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (nvpair_type(elem) != DATA_TYPE_STRING) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be a string"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } (void) nvpair_value_string(elem, &strval); if (strcmp(strval, ZFS_FEATURE_ENABLED) != 0 && strcmp(strval, ZFS_FEATURE_DISABLED) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set to " "'enabled' or 'disabled'"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (!flags.create && strcmp(strval, ZFS_FEATURE_DISABLED) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set to " "'disabled' at creation time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (nvlist_add_uint64(retprops, propname, 0) != 0) { (void) no_memory(hdl); goto error; } continue; } /* * Make sure this property is valid and applies to this type. */ if (prop == ZPOOL_PROP_INVAL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid property '%s'"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (zpool_prop_readonly(prop)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' " "is readonly"), propname); (void) zfs_error(hdl, EZFS_PROPREADONLY, errbuf); goto error; } if (!flags.create && zpool_prop_setonce(prop)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set at " "creation time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (zprop_parse_value(hdl, elem, prop, ZFS_TYPE_POOL, retprops, &strval, &intval, errbuf) != 0) goto error; /* * Perform additional checking for specific properties. */ switch (prop) { case ZPOOL_PROP_VERSION: if (intval < version || !SPA_VERSION_IS_SUPPORTED(intval)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' number %llu is invalid."), propname, (unsigned long long)intval); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); goto error; } break; case ZPOOL_PROP_ASHIFT: if (intval != 0 && (intval < ASHIFT_MIN || intval > ASHIFT_MAX)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' number %llu is invalid, " "only values between %" PRId32 " and %" PRId32 " are allowed."), propname, (unsigned long long)intval, ASHIFT_MIN, ASHIFT_MAX); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_BOOTFS: if (flags.create || flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' cannot be set at creation " "or import time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (version < SPA_VERSION_BOOTFS) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgraded to support " "'%s' property"), propname); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); goto error; } /* * bootfs property value has to be a dataset name and * the dataset has to be in the same pool as it sets to. */ if (!bootfs_name_valid(poolname, strval)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' " "is an invalid name"), strval); (void) zfs_error(hdl, EZFS_INVALIDNAME, errbuf); goto error; } if ((zhp = zpool_open_canfail(hdl, poolname)) == NULL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "could not open pool '%s'"), poolname); (void) zfs_error(hdl, EZFS_OPENFAILED, errbuf); goto error; } zpool_close(zhp); break; case ZPOOL_PROP_ALTROOT: if (!flags.create && !flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set during pool " "creation or import"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (strval[0] != '/') { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "bad alternate root '%s'"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } break; case ZPOOL_PROP_CACHEFILE: if (strval[0] == '\0') break; if (strcmp(strval, "none") == 0) break; if (strval[0] != '/') { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' must be empty, an " "absolute path, or 'none'"), propname); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } slash = strrchr(strval, '/'); if (slash[1] == '\0' || strcmp(slash, "/.") == 0 || strcmp(slash, "/..") == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is not a valid file"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } *slash = '\0'; if (strval[0] != '\0' && (stat64(strval, &statbuf) != 0 || !S_ISDIR(statbuf.st_mode))) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is not a valid directory"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } *slash = '/'; break; case ZPOOL_PROP_COMPATIBILITY: switch (zpool_load_compat(strval, NULL, report, 1024)) { case ZPOOL_COMPATIBILITY_OK: case ZPOOL_COMPATIBILITY_WARNTOKEN: break; case ZPOOL_COMPATIBILITY_BADFILE: case ZPOOL_COMPATIBILITY_BADTOKEN: case ZPOOL_COMPATIBILITY_NOFILES: zfs_error_aux(hdl, "%s", report); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_COMMENT: for (check = strval; *check != '\0'; check++) { if (!isprint(*check)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "comment may only have printable " "characters")); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } } if (strlen(strval) > ZPROP_MAX_COMMENT) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "comment must not exceed %d characters"), ZPROP_MAX_COMMENT); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_READONLY: if (!flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set at " "import time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_MULTIHOST: if (get_system_hostid() == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "requires a non-zero system hostid")); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_DEDUPDITTO: printf("Note: property '%s' no longer has " "any effect\n", propname); break; default: break; } } return (retprops); error: nvlist_free(retprops); return (NULL); } /* * Set zpool property : propname=propval. */ int zpool_set_prop(zpool_handle_t *zhp, const char *propname, const char *propval) { zfs_cmd_t zc = {"\0"}; int ret = -1; char errbuf[ERRBUFLEN]; nvlist_t *nvl = NULL; nvlist_t *realprops; uint64_t version; prop_flags_t flags = { 0 }; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot set property for '%s'"), zhp->zpool_name); if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); if (nvlist_add_string(nvl, propname, propval) != 0) { nvlist_free(nvl); return (no_memory(zhp->zpool_hdl)); } version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if ((realprops = zpool_valid_proplist(zhp->zpool_hdl, zhp->zpool_name, nvl, version, flags, errbuf)) == NULL) { nvlist_free(nvl); return (-1); } nvlist_free(nvl); nvl = realprops; /* * Execute the corresponding ioctl() to set this property. */ (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zcmd_write_src_nvlist(zhp->zpool_hdl, &zc, nvl); ret = zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_SET_PROPS, &zc); zcmd_free_nvlists(&zc); nvlist_free(nvl); if (ret) (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf); else (void) zpool_props_refresh(zhp); return (ret); } int zpool_expand_proplist(zpool_handle_t *zhp, zprop_list_t **plp, zfs_type_t type, boolean_t literal) { libzfs_handle_t *hdl = zhp->zpool_hdl; zprop_list_t *entry; char buf[ZFS_MAXPROPLEN]; nvlist_t *features = NULL; nvpair_t *nvp; zprop_list_t **last; boolean_t firstexpand = (NULL == *plp); int i; if (zprop_expand_list(hdl, plp, type) != 0) return (-1); if (type == ZFS_TYPE_VDEV) return (0); last = plp; while (*last != NULL) last = &(*last)->pl_next; if ((*plp)->pl_all) features = zpool_get_features(zhp); if ((*plp)->pl_all && firstexpand) { for (i = 0; i < SPA_FEATURES; i++) { zprop_list_t *entry = zfs_alloc(hdl, sizeof (zprop_list_t)); entry->pl_prop = ZPROP_USERPROP; entry->pl_user_prop = zfs_asprintf(hdl, "feature@%s", spa_feature_table[i].fi_uname); entry->pl_width = strlen(entry->pl_user_prop); entry->pl_all = B_TRUE; *last = entry; last = &entry->pl_next; } } /* add any unsupported features */ for (nvp = nvlist_next_nvpair(features, NULL); nvp != NULL; nvp = nvlist_next_nvpair(features, nvp)) { char *propname; boolean_t found; zprop_list_t *entry; if (zfeature_is_supported(nvpair_name(nvp))) continue; propname = zfs_asprintf(hdl, "unsupported@%s", nvpair_name(nvp)); /* * Before adding the property to the list make sure that no * other pool already added the same property. */ found = B_FALSE; entry = *plp; while (entry != NULL) { if (entry->pl_user_prop != NULL && strcmp(propname, entry->pl_user_prop) == 0) { found = B_TRUE; break; } entry = entry->pl_next; } if (found) { free(propname); continue; } entry = zfs_alloc(hdl, sizeof (zprop_list_t)); entry->pl_prop = ZPROP_USERPROP; entry->pl_user_prop = propname; entry->pl_width = strlen(entry->pl_user_prop); entry->pl_all = B_TRUE; *last = entry; last = &entry->pl_next; } for (entry = *plp; entry != NULL; entry = entry->pl_next) { if (entry->pl_fixed && !literal) continue; if (entry->pl_prop != ZPROP_USERPROP && zpool_get_prop(zhp, entry->pl_prop, buf, sizeof (buf), NULL, literal) == 0) { if (strlen(buf) > entry->pl_width) entry->pl_width = strlen(buf); } } return (0); } int vdev_expand_proplist(zpool_handle_t *zhp, const char *vdevname, zprop_list_t **plp) { zprop_list_t *entry; char buf[ZFS_MAXPROPLEN]; char *strval = NULL; int err = 0; nvpair_t *elem = NULL; nvlist_t *vprops = NULL; nvlist_t *propval = NULL; const char *propname; vdev_prop_t prop; zprop_list_t **last; for (entry = *plp; entry != NULL; entry = entry->pl_next) { if (entry->pl_fixed) continue; if (zpool_get_vdev_prop(zhp, vdevname, entry->pl_prop, entry->pl_user_prop, buf, sizeof (buf), NULL, B_FALSE) == 0) { if (strlen(buf) > entry->pl_width) entry->pl_width = strlen(buf); } if (entry->pl_prop == VDEV_PROP_NAME && strlen(vdevname) > entry->pl_width) entry->pl_width = strlen(vdevname); } /* Handle the all properties case */ last = plp; if (*last != NULL && (*last)->pl_all == B_TRUE) { while (*last != NULL) last = &(*last)->pl_next; err = zpool_get_all_vdev_props(zhp, vdevname, &vprops); if (err != 0) return (err); while ((elem = nvlist_next_nvpair(vprops, elem)) != NULL) { propname = nvpair_name(elem); /* Skip properties that are not user defined */ if ((prop = vdev_name_to_prop(propname)) != VDEV_PROP_USERPROP) continue; if (nvpair_value_nvlist(elem, &propval) != 0) continue; strval = fnvlist_lookup_string(propval, ZPROP_VALUE); entry = zfs_alloc(zhp->zpool_hdl, sizeof (zprop_list_t)); entry->pl_prop = prop; entry->pl_user_prop = zfs_strdup(zhp->zpool_hdl, propname); entry->pl_width = strlen(strval); entry->pl_all = B_TRUE; *last = entry; last = &entry->pl_next; } } return (0); } /* * Get the state for the given feature on the given ZFS pool. */ int zpool_prop_get_feature(zpool_handle_t *zhp, const char *propname, char *buf, size_t len) { uint64_t refcount; boolean_t found = B_FALSE; nvlist_t *features = zpool_get_features(zhp); boolean_t supported; const char *feature = strchr(propname, '@') + 1; supported = zpool_prop_feature(propname); ASSERT(supported || zpool_prop_unsupported(propname)); /* * Convert from feature name to feature guid. This conversion is * unnecessary for unsupported@... properties because they already * use guids. */ if (supported) { int ret; spa_feature_t fid; ret = zfeature_lookup_name(feature, &fid); if (ret != 0) { (void) strlcpy(buf, "-", len); return (ENOTSUP); } feature = spa_feature_table[fid].fi_guid; } if (nvlist_lookup_uint64(features, feature, &refcount) == 0) found = B_TRUE; if (supported) { if (!found) { (void) strlcpy(buf, ZFS_FEATURE_DISABLED, len); } else { if (refcount == 0) (void) strlcpy(buf, ZFS_FEATURE_ENABLED, len); else (void) strlcpy(buf, ZFS_FEATURE_ACTIVE, len); } } else { if (found) { if (refcount == 0) { (void) strcpy(buf, ZFS_UNSUPPORTED_INACTIVE); } else { (void) strcpy(buf, ZFS_UNSUPPORTED_READONLY); } } else { (void) strlcpy(buf, "-", len); return (ENOTSUP); } } return (0); } /* * Validate the given pool name, optionally putting an extended error message in * 'buf'. */ boolean_t zpool_name_valid(libzfs_handle_t *hdl, boolean_t isopen, const char *pool) { namecheck_err_t why; char what; int ret; ret = pool_namecheck(pool, &why, &what); /* * The rules for reserved pool names were extended at a later point. * But we need to support users with existing pools that may now be * invalid. So we only check for this expanded set of names during a * create (or import), and only in userland. */ if (ret == 0 && !isopen && (strncmp(pool, "mirror", 6) == 0 || strncmp(pool, "raidz", 5) == 0 || strncmp(pool, "draid", 5) == 0 || strncmp(pool, "spare", 5) == 0 || strcmp(pool, "log") == 0)) { if (hdl != NULL) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is reserved")); return (B_FALSE); } if (ret != 0) { if (hdl != NULL) { switch (why) { case NAME_ERR_TOOLONG: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is too long")); break; case NAME_ERR_INVALCHAR: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid character " "'%c' in pool name"), what); break; case NAME_ERR_NOLETTER: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name must begin with a letter")); break; case NAME_ERR_RESERVED: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is reserved")); break; case NAME_ERR_DISKLIKE: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool name is reserved")); break; case NAME_ERR_LEADING_SLASH: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "leading slash in name")); break; case NAME_ERR_EMPTY_COMPONENT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "empty component in name")); break; case NAME_ERR_TRAILING_SLASH: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "trailing slash in name")); break; case NAME_ERR_MULTIPLE_DELIMITERS: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "multiple '@' and/or '#' delimiters in " "name")); break; case NAME_ERR_NO_AT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "permission set is missing '@'")); break; default: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "(%d) not defined"), why); break; } } return (B_FALSE); } return (B_TRUE); } /* * Open a handle to the given pool, even if the pool is currently in the FAULTED * state. */ zpool_handle_t * zpool_open_canfail(libzfs_handle_t *hdl, const char *pool) { zpool_handle_t *zhp; boolean_t missing; /* * Make sure the pool name is valid. */ if (!zpool_name_valid(hdl, B_TRUE, pool)) { (void) zfs_error_fmt(hdl, EZFS_INVALIDNAME, dgettext(TEXT_DOMAIN, "cannot open '%s'"), pool); return (NULL); } zhp = zfs_alloc(hdl, sizeof (zpool_handle_t)); zhp->zpool_hdl = hdl; (void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name)); if (zpool_refresh_stats(zhp, &missing) != 0) { zpool_close(zhp); return (NULL); } if (missing) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "no such pool")); (void) zfs_error_fmt(hdl, EZFS_NOENT, dgettext(TEXT_DOMAIN, "cannot open '%s'"), pool); zpool_close(zhp); return (NULL); } return (zhp); } /* * Like the above, but silent on error. Used when iterating over pools (because * the configuration cache may be out of date). */ int zpool_open_silent(libzfs_handle_t *hdl, const char *pool, zpool_handle_t **ret) { zpool_handle_t *zhp; boolean_t missing; zhp = zfs_alloc(hdl, sizeof (zpool_handle_t)); zhp->zpool_hdl = hdl; (void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name)); if (zpool_refresh_stats(zhp, &missing) != 0) { zpool_close(zhp); return (-1); } if (missing) { zpool_close(zhp); *ret = NULL; return (0); } *ret = zhp; return (0); } /* * Similar to zpool_open_canfail(), but refuses to open pools in the faulted * state. */ zpool_handle_t * zpool_open(libzfs_handle_t *hdl, const char *pool) { zpool_handle_t *zhp; if ((zhp = zpool_open_canfail(hdl, pool)) == NULL) return (NULL); if (zhp->zpool_state == POOL_STATE_UNAVAIL) { (void) zfs_error_fmt(hdl, EZFS_POOLUNAVAIL, dgettext(TEXT_DOMAIN, "cannot open '%s'"), zhp->zpool_name); zpool_close(zhp); return (NULL); } return (zhp); } /* * Close the handle. Simply frees the memory associated with the handle. */ void zpool_close(zpool_handle_t *zhp) { nvlist_free(zhp->zpool_config); nvlist_free(zhp->zpool_old_config); nvlist_free(zhp->zpool_props); free(zhp); } /* * Return the name of the pool. */ const char * zpool_get_name(zpool_handle_t *zhp) { return (zhp->zpool_name); } /* * Return the state of the pool (ACTIVE or UNAVAILABLE) */ int zpool_get_state(zpool_handle_t *zhp) { return (zhp->zpool_state); } /* * Check if vdev list contains a special vdev */ static boolean_t zpool_has_special_vdev(nvlist_t *nvroot) { nvlist_t **child; uint_t children; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { for (uint_t c = 0; c < children; c++) { char *bias; if (nvlist_lookup_string(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0 && strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) { return (B_TRUE); } } } return (B_FALSE); } /* * Check if vdev list contains a dRAID vdev */ static boolean_t zpool_has_draid_vdev(nvlist_t *nvroot) { nvlist_t **child; uint_t children; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { for (uint_t c = 0; c < children; c++) { char *type; if (nvlist_lookup_string(child[c], ZPOOL_CONFIG_TYPE, &type) == 0 && strcmp(type, VDEV_TYPE_DRAID) == 0) { return (B_TRUE); } } } return (B_FALSE); } /* * Output a dRAID top-level vdev name in to the provided buffer. */ static char * zpool_draid_name(char *name, int len, uint64_t data, uint64_t parity, uint64_t spares, uint64_t children) { snprintf(name, len, "%s%llu:%llud:%lluc:%llus", VDEV_TYPE_DRAID, (u_longlong_t)parity, (u_longlong_t)data, (u_longlong_t)children, (u_longlong_t)spares); return (name); } /* * Return B_TRUE if the provided name is a dRAID spare name. */ boolean_t zpool_is_draid_spare(const char *name) { uint64_t spare_id, parity, vdev_id; if (sscanf(name, VDEV_TYPE_DRAID "%llu-%llu-%llu", (u_longlong_t *)&parity, (u_longlong_t *)&vdev_id, (u_longlong_t *)&spare_id) == 3) { return (B_TRUE); } return (B_FALSE); } /* * Create the named pool, using the provided vdev list. It is assumed * that the consumer has already validated the contents of the nvlist, so we * don't have to worry about error semantics. */ int zpool_create(libzfs_handle_t *hdl, const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *fsprops) { zfs_cmd_t zc = {"\0"}; nvlist_t *zc_fsprops = NULL; nvlist_t *zc_props = NULL; nvlist_t *hidden_args = NULL; uint8_t *wkeydata = NULL; uint_t wkeylen = 0; char errbuf[ERRBUFLEN]; int ret = -1; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot create '%s'"), pool); if (!zpool_name_valid(hdl, B_FALSE, pool)) return (zfs_error(hdl, EZFS_INVALIDNAME, errbuf)); zcmd_write_conf_nvlist(hdl, &zc, nvroot); if (props) { prop_flags_t flags = { .create = B_TRUE, .import = B_FALSE }; if ((zc_props = zpool_valid_proplist(hdl, pool, props, SPA_VERSION_1, flags, errbuf)) == NULL) { goto create_failed; } } if (fsprops) { uint64_t zoned; char *zonestr; zoned = ((nvlist_lookup_string(fsprops, zfs_prop_to_name(ZFS_PROP_ZONED), &zonestr) == 0) && strcmp(zonestr, "on") == 0); if ((zc_fsprops = zfs_valid_proplist(hdl, ZFS_TYPE_FILESYSTEM, fsprops, zoned, NULL, NULL, B_TRUE, errbuf)) == NULL) { goto create_failed; } if (nvlist_exists(zc_fsprops, zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS)) && !zpool_has_special_vdev(nvroot)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s property requires a special vdev"), zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS)); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto create_failed; } if (!zc_props && (nvlist_alloc(&zc_props, NV_UNIQUE_NAME, 0) != 0)) { goto create_failed; } if (zfs_crypto_create(hdl, NULL, zc_fsprops, props, B_TRUE, &wkeydata, &wkeylen) != 0) { zfs_error(hdl, EZFS_CRYPTOFAILED, errbuf); goto create_failed; } if (nvlist_add_nvlist(zc_props, ZPOOL_ROOTFS_PROPS, zc_fsprops) != 0) { goto create_failed; } if (wkeydata != NULL) { if (nvlist_alloc(&hidden_args, NV_UNIQUE_NAME, 0) != 0) goto create_failed; if (nvlist_add_uint8_array(hidden_args, "wkeydata", wkeydata, wkeylen) != 0) goto create_failed; if (nvlist_add_nvlist(zc_props, ZPOOL_HIDDEN_ARGS, hidden_args) != 0) goto create_failed; } } if (zc_props) zcmd_write_src_nvlist(hdl, &zc, zc_props); (void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name)); if ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_CREATE, &zc)) != 0) { zcmd_free_nvlists(&zc); nvlist_free(zc_props); nvlist_free(zc_fsprops); nvlist_free(hidden_args); if (wkeydata != NULL) free(wkeydata); switch (errno) { case EBUSY: /* * This can happen if the user has specified the same * device multiple times. We can't reliably detect this * until we try to add it and see we already have a * label. This can also happen under if the device is * part of an active md or lvm device. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more vdevs refer to the same device, or " "one of\nthe devices is part of an active md or " "lvm device")); return (zfs_error(hdl, EZFS_BADDEV, errbuf)); case ERANGE: /* * This happens if the record size is smaller or larger * than the allowed size range, or not a power of 2. * * NOTE: although zfs_valid_proplist is called earlier, * this case may have slipped through since the * pool does not exist yet and it is therefore * impossible to read properties e.g. max blocksize * from the pool. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "record size invalid")); return (zfs_error(hdl, EZFS_BADPROP, errbuf)); case EOVERFLOW: /* * This occurs when one of the devices is below * SPA_MINDEVSIZE. Unfortunately, we can't detect which * device was the problem device since there's no * reliable way to determine device size from userland. */ { char buf[64]; zfs_nicebytes(SPA_MINDEVSIZE, buf, sizeof (buf)); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is less than the " "minimum size (%s)"), buf); } return (zfs_error(hdl, EZFS_BADDEV, errbuf)); case ENOSPC: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is out of space")); return (zfs_error(hdl, EZFS_BADDEV, errbuf)); case EINVAL: if (zpool_has_draid_vdev(nvroot) && zfeature_lookup_name("draid", NULL) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dRAID vdevs are unsupported by the " "kernel")); return (zfs_error(hdl, EZFS_BADDEV, errbuf)); } else { return (zpool_standard_error(hdl, errno, errbuf)); } default: return (zpool_standard_error(hdl, errno, errbuf)); } } create_failed: zcmd_free_nvlists(&zc); nvlist_free(zc_props); nvlist_free(zc_fsprops); nvlist_free(hidden_args); if (wkeydata != NULL) free(wkeydata); return (ret); } /* * Destroy the given pool. It is up to the caller to ensure that there are no * datasets left in the pool. */ int zpool_destroy(zpool_handle_t *zhp, const char *log_str) { zfs_cmd_t zc = {"\0"}; zfs_handle_t *zfp = NULL; libzfs_handle_t *hdl = zhp->zpool_hdl; char errbuf[ERRBUFLEN]; if (zhp->zpool_state == POOL_STATE_ACTIVE && (zfp = zfs_open(hdl, zhp->zpool_name, ZFS_TYPE_FILESYSTEM)) == NULL) return (-1); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(hdl, ZFS_IOC_POOL_DESTROY, &zc) != 0) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot destroy '%s'"), zhp->zpool_name); if (errno == EROFS) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is read only")); (void) zfs_error(hdl, EZFS_BADDEV, errbuf); } else { (void) zpool_standard_error(hdl, errno, errbuf); } if (zfp) zfs_close(zfp); return (-1); } if (zfp) { remove_mountpoint(zfp); zfs_close(zfp); } return (0); } /* * Create a checkpoint in the given pool. */ int zpool_checkpoint(zpool_handle_t *zhp) { libzfs_handle_t *hdl = zhp->zpool_hdl; char errbuf[ERRBUFLEN]; int error; error = lzc_pool_checkpoint(zhp->zpool_name); if (error != 0) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot checkpoint '%s'"), zhp->zpool_name); (void) zpool_standard_error(hdl, error, errbuf); return (-1); } return (0); } /* * Discard the checkpoint from the given pool. */ int zpool_discard_checkpoint(zpool_handle_t *zhp) { libzfs_handle_t *hdl = zhp->zpool_hdl; char errbuf[ERRBUFLEN]; int error; error = lzc_pool_checkpoint_discard(zhp->zpool_name); if (error != 0) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot discard checkpoint in '%s'"), zhp->zpool_name); (void) zpool_standard_error(hdl, error, errbuf); return (-1); } return (0); } /* * Add the given vdevs to the pool. The caller must have already performed the * necessary verification to ensure that the vdev specification is well-formed. */ int zpool_add(zpool_handle_t *zhp, nvlist_t *nvroot) { zfs_cmd_t zc = {"\0"}; int ret; libzfs_handle_t *hdl = zhp->zpool_hdl; char errbuf[ERRBUFLEN]; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot add to '%s'"), zhp->zpool_name); if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) < SPA_VERSION_SPARES && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be " "upgraded to add hot spares")); return (zfs_error(hdl, EZFS_BADVERSION, errbuf)); } if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) < SPA_VERSION_L2CACHE && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be " "upgraded to add cache devices")); return (zfs_error(hdl, EZFS_BADVERSION, errbuf)); } zcmd_write_conf_nvlist(hdl, &zc, nvroot); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_ADD, &zc) != 0) { switch (errno) { case EBUSY: /* * This can happen if the user has specified the same * device multiple times. We can't reliably detect this * until we try to add it and see we already have a * label. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more vdevs refer to the same device")); (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case EINVAL: if (zpool_has_draid_vdev(nvroot) && zfeature_lookup_name("draid", NULL) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dRAID vdevs are unsupported by the " "kernel")); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid config; a pool with removing/" "removed vdevs does not support adding " "raidz or dRAID vdevs")); } (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case EOVERFLOW: /* * This occurs when one of the devices is below * SPA_MINDEVSIZE. Unfortunately, we can't detect which * device was the problem device since there's no * reliable way to determine device size from userland. */ { char buf[64]; zfs_nicebytes(SPA_MINDEVSIZE, buf, sizeof (buf)); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device is less than the minimum " "size (%s)"), buf); } (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case ENOTSUP: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgraded to add these vdevs")); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); break; default: (void) zpool_standard_error(hdl, errno, errbuf); } ret = -1; } else { ret = 0; } zcmd_free_nvlists(&zc); return (ret); } /* * Exports the pool from the system. The caller must ensure that there are no * mounted datasets in the pool. */ static int zpool_export_common(zpool_handle_t *zhp, boolean_t force, boolean_t hardforce, const char *log_str) { zfs_cmd_t zc = {"\0"}; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = force; zc.zc_guid = hardforce; zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_EXPORT, &zc) != 0) { switch (errno) { case EXDEV: zfs_error_aux(zhp->zpool_hdl, dgettext(TEXT_DOMAIN, "use '-f' to override the following errors:\n" "'%s' has an active shared spare which could be" " used by other pools once '%s' is exported."), zhp->zpool_name, zhp->zpool_name); return (zfs_error_fmt(zhp->zpool_hdl, EZFS_ACTIVE_SPARE, dgettext(TEXT_DOMAIN, "cannot export '%s'"), zhp->zpool_name)); default: return (zpool_standard_error_fmt(zhp->zpool_hdl, errno, dgettext(TEXT_DOMAIN, "cannot export '%s'"), zhp->zpool_name)); } } return (0); } int zpool_export(zpool_handle_t *zhp, boolean_t force, const char *log_str) { return (zpool_export_common(zhp, force, B_FALSE, log_str)); } int zpool_export_force(zpool_handle_t *zhp, const char *log_str) { return (zpool_export_common(zhp, B_TRUE, B_TRUE, log_str)); } static void zpool_rewind_exclaim(libzfs_handle_t *hdl, const char *name, boolean_t dryrun, nvlist_t *config) { nvlist_t *nv = NULL; uint64_t rewindto; int64_t loss = -1; struct tm t; char timestr[128]; if (!hdl->libzfs_printerr || config == NULL) return; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 || nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0) { return; } if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0) return; (void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss); if (localtime_r((time_t *)&rewindto, &t) != NULL && strftime(timestr, 128, "%c", &t) != 0) { if (dryrun) { (void) printf(dgettext(TEXT_DOMAIN, "Would be able to return %s " "to its state as of %s.\n"), name, timestr); } else { (void) printf(dgettext(TEXT_DOMAIN, "Pool %s returned to its state as of %s.\n"), name, timestr); } if (loss > 120) { (void) printf(dgettext(TEXT_DOMAIN, "%s approximately %lld "), dryrun ? "Would discard" : "Discarded", ((longlong_t)loss + 30) / 60); (void) printf(dgettext(TEXT_DOMAIN, "minutes of transactions.\n")); } else if (loss > 0) { (void) printf(dgettext(TEXT_DOMAIN, "%s approximately %lld "), dryrun ? "Would discard" : "Discarded", (longlong_t)loss); (void) printf(dgettext(TEXT_DOMAIN, "seconds of transactions.\n")); } } } void zpool_explain_recover(libzfs_handle_t *hdl, const char *name, int reason, nvlist_t *config) { nvlist_t *nv = NULL; int64_t loss = -1; uint64_t edata = UINT64_MAX; uint64_t rewindto; struct tm t; char timestr[128]; if (!hdl->libzfs_printerr) return; if (reason >= 0) (void) printf(dgettext(TEXT_DOMAIN, "action: ")); else (void) printf(dgettext(TEXT_DOMAIN, "\t")); /* All attempted rewinds failed if ZPOOL_CONFIG_LOAD_TIME missing */ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 || nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0 || nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0) goto no_info; (void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_DATA_ERRORS, &edata); (void) printf(dgettext(TEXT_DOMAIN, "Recovery is possible, but will result in some data loss.\n")); if (localtime_r((time_t *)&rewindto, &t) != NULL && strftime(timestr, 128, "%c", &t) != 0) { (void) printf(dgettext(TEXT_DOMAIN, "\tReturning the pool to its state as of %s\n" "\tshould correct the problem. "), timestr); } else { (void) printf(dgettext(TEXT_DOMAIN, "\tReverting the pool to an earlier state " "should correct the problem.\n\t")); } if (loss > 120) { (void) printf(dgettext(TEXT_DOMAIN, "Approximately %lld minutes of data\n" "\tmust be discarded, irreversibly. "), ((longlong_t)loss + 30) / 60); } else if (loss > 0) { (void) printf(dgettext(TEXT_DOMAIN, "Approximately %lld seconds of data\n" "\tmust be discarded, irreversibly. "), (longlong_t)loss); } if (edata != 0 && edata != UINT64_MAX) { if (edata == 1) { (void) printf(dgettext(TEXT_DOMAIN, "After rewind, at least\n" "\tone persistent user-data error will remain. ")); } else { (void) printf(dgettext(TEXT_DOMAIN, "After rewind, several\n" "\tpersistent user-data errors will remain. ")); } } (void) printf(dgettext(TEXT_DOMAIN, "Recovery can be attempted\n\tby executing 'zpool %s -F %s'. "), reason >= 0 ? "clear" : "import", name); (void) printf(dgettext(TEXT_DOMAIN, "A scrub of the pool\n" "\tis strongly recommended after recovery.\n")); return; no_info: (void) printf(dgettext(TEXT_DOMAIN, "Destroy and re-create the pool from\n\ta backup source.\n")); } /* * zpool_import() is a contracted interface. Should be kept the same * if possible. * * Applications should use zpool_import_props() to import a pool with * new properties value to be set. */ int zpool_import(libzfs_handle_t *hdl, nvlist_t *config, const char *newname, char *altroot) { nvlist_t *props = NULL; int ret; if (altroot != NULL) { if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) { return (zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); } if (nvlist_add_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), altroot) != 0 || nvlist_add_string(props, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), "none") != 0) { nvlist_free(props); return (zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); } } ret = zpool_import_props(hdl, config, newname, props, ZFS_IMPORT_NORMAL); nvlist_free(props); return (ret); } static void print_vdev_tree(libzfs_handle_t *hdl, const char *name, nvlist_t *nv, int indent) { nvlist_t **child; uint_t c, children; char *vname; uint64_t is_log = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log); if (name != NULL) (void) printf("\t%*s%s%s\n", indent, "", name, is_log ? " [log]" : ""); if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return; for (c = 0; c < children; c++) { vname = zpool_vdev_name(hdl, NULL, child[c], VDEV_NAME_TYPE_ID); print_vdev_tree(hdl, vname, child[c], indent + 2); free(vname); } } void zpool_print_unsup_feat(nvlist_t *config) { nvlist_t *nvinfo, *unsup_feat; nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); unsup_feat = fnvlist_lookup_nvlist(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT); for (nvpair_t *nvp = nvlist_next_nvpair(unsup_feat, NULL); nvp != NULL; nvp = nvlist_next_nvpair(unsup_feat, nvp)) { char *desc = fnvpair_value_string(nvp); if (strlen(desc) > 0) (void) printf("\t%s (%s)\n", nvpair_name(nvp), desc); else (void) printf("\t%s\n", nvpair_name(nvp)); } } /* * Import the given pool using the known configuration and a list of * properties to be set. The configuration should have come from * zpool_find_import(). The 'newname' parameters control whether the pool * is imported with a different name. */ int zpool_import_props(libzfs_handle_t *hdl, nvlist_t *config, const char *newname, nvlist_t *props, int flags) { zfs_cmd_t zc = {"\0"}; zpool_load_policy_t policy; nvlist_t *nv = NULL; nvlist_t *nvinfo = NULL; nvlist_t *missing = NULL; const char *thename; char *origname; int ret; int error = 0; char errbuf[ERRBUFLEN]; origname = fnvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME); (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot import pool '%s'"), origname); if (newname != NULL) { if (!zpool_name_valid(hdl, B_FALSE, newname)) return (zfs_error_fmt(hdl, EZFS_INVALIDNAME, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); thename = newname; } else { thename = origname; } if (props != NULL) { uint64_t version; prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE }; version = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION); if ((props = zpool_valid_proplist(hdl, origname, props, version, flags, errbuf)) == NULL) return (-1); zcmd_write_src_nvlist(hdl, &zc, props); nvlist_free(props); } (void) strlcpy(zc.zc_name, thename, sizeof (zc.zc_name)); zc.zc_guid = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID); zcmd_write_conf_nvlist(hdl, &zc, config); zcmd_alloc_dst_nvlist(hdl, &zc, zc.zc_nvlist_conf_size * 2); zc.zc_cookie = flags; while ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_IMPORT, &zc)) != 0 && errno == ENOMEM) zcmd_expand_dst_nvlist(hdl, &zc); if (ret != 0) error = errno; (void) zcmd_read_dst_nvlist(hdl, &zc, &nv); zcmd_free_nvlists(&zc); zpool_get_load_policy(config, &policy); if (error) { char desc[1024]; char aux[256]; /* * Dry-run failed, but we print out what success * looks like if we found a best txg */ if (policy.zlp_rewind & ZPOOL_TRY_REWIND) { zpool_rewind_exclaim(hdl, newname ? origname : thename, B_TRUE, nv); nvlist_free(nv); return (-1); } if (newname == NULL) (void) snprintf(desc, sizeof (desc), dgettext(TEXT_DOMAIN, "cannot import '%s'"), thename); else (void) snprintf(desc, sizeof (desc), dgettext(TEXT_DOMAIN, "cannot import '%s' as '%s'"), origname, thename); switch (error) { case ENOTSUP: if (nv != NULL && nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 && nvlist_exists(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT)) { (void) printf(dgettext(TEXT_DOMAIN, "This " "pool uses the following feature(s) not " "supported by this system:\n")); zpool_print_unsup_feat(nv); if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY)) { (void) printf(dgettext(TEXT_DOMAIN, "All unsupported features are only " "required for writing to the pool." "\nThe pool can be imported using " "'-o readonly=on'.\n")); } } /* * Unsupported version. */ (void) zfs_error(hdl, EZFS_BADVERSION, desc); break; case EREMOTEIO: if (nv != NULL && nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0) { const char *hostname = ""; uint64_t hostid = 0; mmp_state_t mmp_state; mmp_state = fnvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_STATE); if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTNAME)) hostname = fnvlist_lookup_string(nvinfo, ZPOOL_CONFIG_MMP_HOSTNAME); if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTID)) hostid = fnvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_HOSTID); if (mmp_state == MMP_STATE_ACTIVE) { (void) snprintf(aux, sizeof (aux), dgettext(TEXT_DOMAIN, "pool is imp" "orted on host '%s' (hostid=%lx).\n" "Export the pool on the other " "system, then run 'zpool import'."), hostname, (unsigned long) hostid); } else if (mmp_state == MMP_STATE_NO_HOSTID) { (void) snprintf(aux, sizeof (aux), dgettext(TEXT_DOMAIN, "pool has " "the multihost property on and " "the\nsystem's hostid is not set. " "Set a unique system hostid with " "the zgenhostid(8) command.\n")); } (void) zfs_error_aux(hdl, "%s", aux); } (void) zfs_error(hdl, EZFS_ACTIVE_POOL, desc); break; case EINVAL: (void) zfs_error(hdl, EZFS_INVALCONFIG, desc); break; case EROFS: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is read only")); (void) zfs_error(hdl, EZFS_BADDEV, desc); break; case ENXIO: if (nv && nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 && nvlist_lookup_nvlist(nvinfo, ZPOOL_CONFIG_MISSING_DEVICES, &missing) == 0) { (void) printf(dgettext(TEXT_DOMAIN, "The devices below are missing or " "corrupted, use '-m' to import the pool " "anyway:\n")); print_vdev_tree(hdl, NULL, missing, 2); (void) printf("\n"); } (void) zpool_standard_error(hdl, error, desc); break; case EEXIST: (void) zpool_standard_error(hdl, error, desc); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices are already in use\n")); (void) zfs_error(hdl, EZFS_BADDEV, desc); break; case ENAMETOOLONG: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new name of at least one dataset is longer than " "the maximum allowable length")); (void) zfs_error(hdl, EZFS_NAMETOOLONG, desc); break; default: (void) zpool_standard_error(hdl, error, desc); zpool_explain_recover(hdl, newname ? origname : thename, -error, nv); break; } nvlist_free(nv); ret = -1; } else { zpool_handle_t *zhp; /* * This should never fail, but play it safe anyway. */ if (zpool_open_silent(hdl, thename, &zhp) != 0) ret = -1; else if (zhp != NULL) zpool_close(zhp); if (policy.zlp_rewind & (ZPOOL_DO_REWIND | ZPOOL_TRY_REWIND)) { zpool_rewind_exclaim(hdl, newname ? origname : thename, ((policy.zlp_rewind & ZPOOL_TRY_REWIND) != 0), nv); } nvlist_free(nv); return (0); } return (ret); } /* * Translate vdev names to guids. If a vdev_path is determined to be * unsuitable then a vd_errlist is allocated and the vdev path and errno * are added to it. */ static int zpool_translate_vdev_guids(zpool_handle_t *zhp, nvlist_t *vds, nvlist_t *vdev_guids, nvlist_t *guids_to_paths, nvlist_t **vd_errlist) { nvlist_t *errlist = NULL; int error = 0; for (nvpair_t *elem = nvlist_next_nvpair(vds, NULL); elem != NULL; elem = nvlist_next_nvpair(vds, elem)) { boolean_t spare, cache; char *vd_path = nvpair_name(elem); nvlist_t *tgt = zpool_find_vdev(zhp, vd_path, &spare, &cache, NULL); if ((tgt == NULL) || cache || spare) { if (errlist == NULL) { errlist = fnvlist_alloc(); error = EINVAL; } uint64_t err = (tgt == NULL) ? EZFS_NODEVICE : (spare ? EZFS_ISSPARE : EZFS_ISL2CACHE); fnvlist_add_int64(errlist, vd_path, err); continue; } uint64_t guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); fnvlist_add_uint64(vdev_guids, vd_path, guid); char msg[MAXNAMELEN]; (void) snprintf(msg, sizeof (msg), "%llu", (u_longlong_t)guid); fnvlist_add_string(guids_to_paths, msg, vd_path); } if (error != 0) { verify(errlist != NULL); if (vd_errlist != NULL) *vd_errlist = errlist; else fnvlist_free(errlist); } return (error); } static int xlate_init_err(int err) { switch (err) { case ENODEV: return (EZFS_NODEVICE); case EINVAL: case EROFS: return (EZFS_BADDEV); case EBUSY: return (EZFS_INITIALIZING); case ESRCH: return (EZFS_NO_INITIALIZE); } return (err); } /* * Begin, suspend, or cancel the initialization (initializing of all free * blocks) for the given vdevs in the given pool. */ static int zpool_initialize_impl(zpool_handle_t *zhp, pool_initialize_func_t cmd_type, nvlist_t *vds, boolean_t wait) { int err; nvlist_t *vdev_guids = fnvlist_alloc(); nvlist_t *guids_to_paths = fnvlist_alloc(); nvlist_t *vd_errlist = NULL; nvlist_t *errlist; nvpair_t *elem; err = zpool_translate_vdev_guids(zhp, vds, vdev_guids, guids_to_paths, &vd_errlist); if (err != 0) { verify(vd_errlist != NULL); goto list_errors; } err = lzc_initialize(zhp->zpool_name, cmd_type, vdev_guids, &errlist); if (err != 0) { if (errlist != NULL) { vd_errlist = fnvlist_lookup_nvlist(errlist, ZPOOL_INITIALIZE_VDEVS); goto list_errors; } (void) zpool_standard_error(zhp->zpool_hdl, err, dgettext(TEXT_DOMAIN, "operation failed")); goto out; } if (wait) { for (elem = nvlist_next_nvpair(vdev_guids, NULL); elem != NULL; elem = nvlist_next_nvpair(vdev_guids, elem)) { uint64_t guid = fnvpair_value_uint64(elem); err = lzc_wait_tag(zhp->zpool_name, ZPOOL_WAIT_INITIALIZE, guid, NULL); if (err != 0) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, err, dgettext(TEXT_DOMAIN, "error " "waiting for '%s' to initialize"), nvpair_name(elem)); goto out; } } } goto out; list_errors: for (elem = nvlist_next_nvpair(vd_errlist, NULL); elem != NULL; elem = nvlist_next_nvpair(vd_errlist, elem)) { int64_t vd_error = xlate_init_err(fnvpair_value_int64(elem)); char *path; if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem), &path) != 0) path = nvpair_name(elem); (void) zfs_error_fmt(zhp->zpool_hdl, vd_error, "cannot initialize '%s'", path); } out: fnvlist_free(vdev_guids); fnvlist_free(guids_to_paths); if (vd_errlist != NULL) fnvlist_free(vd_errlist); return (err == 0 ? 0 : -1); } int zpool_initialize(zpool_handle_t *zhp, pool_initialize_func_t cmd_type, nvlist_t *vds) { return (zpool_initialize_impl(zhp, cmd_type, vds, B_FALSE)); } int zpool_initialize_wait(zpool_handle_t *zhp, pool_initialize_func_t cmd_type, nvlist_t *vds) { return (zpool_initialize_impl(zhp, cmd_type, vds, B_TRUE)); } static int xlate_trim_err(int err) { switch (err) { case ENODEV: return (EZFS_NODEVICE); case EINVAL: case EROFS: return (EZFS_BADDEV); case EBUSY: return (EZFS_TRIMMING); case ESRCH: return (EZFS_NO_TRIM); case EOPNOTSUPP: return (EZFS_TRIM_NOTSUP); } return (err); } static int zpool_trim_wait(zpool_handle_t *zhp, nvlist_t *vdev_guids) { int err; nvpair_t *elem; for (elem = nvlist_next_nvpair(vdev_guids, NULL); elem != NULL; elem = nvlist_next_nvpair(vdev_guids, elem)) { uint64_t guid = fnvpair_value_uint64(elem); err = lzc_wait_tag(zhp->zpool_name, ZPOOL_WAIT_TRIM, guid, NULL); if (err != 0) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, err, dgettext(TEXT_DOMAIN, "error " "waiting to trim '%s'"), nvpair_name(elem)); return (err); } } return (0); } /* * Check errlist and report any errors, omitting ones which should be * suppressed. Returns B_TRUE if any errors were reported. */ static boolean_t check_trim_errs(zpool_handle_t *zhp, trimflags_t *trim_flags, nvlist_t *guids_to_paths, nvlist_t *vds, nvlist_t *errlist) { nvpair_t *elem; boolean_t reported_errs = B_FALSE; int num_vds = 0; int num_suppressed_errs = 0; for (elem = nvlist_next_nvpair(vds, NULL); elem != NULL; elem = nvlist_next_nvpair(vds, elem)) { num_vds++; } for (elem = nvlist_next_nvpair(errlist, NULL); elem != NULL; elem = nvlist_next_nvpair(errlist, elem)) { int64_t vd_error = xlate_trim_err(fnvpair_value_int64(elem)); char *path; /* * If only the pool was specified, and it was not a secure * trim then suppress warnings for individual vdevs which * do not support trimming. */ if (vd_error == EZFS_TRIM_NOTSUP && trim_flags->fullpool && !trim_flags->secure) { num_suppressed_errs++; continue; } reported_errs = B_TRUE; if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem), &path) != 0) path = nvpair_name(elem); (void) zfs_error_fmt(zhp->zpool_hdl, vd_error, "cannot trim '%s'", path); } if (num_suppressed_errs == num_vds) { (void) zfs_error_aux(zhp->zpool_hdl, dgettext(TEXT_DOMAIN, "no devices in pool support trim operations")); (void) (zfs_error(zhp->zpool_hdl, EZFS_TRIM_NOTSUP, dgettext(TEXT_DOMAIN, "cannot trim"))); reported_errs = B_TRUE; } return (reported_errs); } /* * Begin, suspend, or cancel the TRIM (discarding of all free blocks) for * the given vdevs in the given pool. */ int zpool_trim(zpool_handle_t *zhp, pool_trim_func_t cmd_type, nvlist_t *vds, trimflags_t *trim_flags) { int err; int retval = 0; nvlist_t *vdev_guids = fnvlist_alloc(); nvlist_t *guids_to_paths = fnvlist_alloc(); nvlist_t *errlist = NULL; err = zpool_translate_vdev_guids(zhp, vds, vdev_guids, guids_to_paths, &errlist); if (err != 0) { check_trim_errs(zhp, trim_flags, guids_to_paths, vds, errlist); retval = -1; goto out; } err = lzc_trim(zhp->zpool_name, cmd_type, trim_flags->rate, trim_flags->secure, vdev_guids, &errlist); if (err != 0) { nvlist_t *vd_errlist; if (errlist != NULL && nvlist_lookup_nvlist(errlist, ZPOOL_TRIM_VDEVS, &vd_errlist) == 0) { if (check_trim_errs(zhp, trim_flags, guids_to_paths, vds, vd_errlist)) { retval = -1; goto out; } } else { char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "operation failed")); zpool_standard_error(zhp->zpool_hdl, err, errbuf); retval = -1; goto out; } } if (trim_flags->wait) retval = zpool_trim_wait(zhp, vdev_guids); out: if (errlist != NULL) fnvlist_free(errlist); fnvlist_free(vdev_guids); fnvlist_free(guids_to_paths); return (retval); } /* * Scan the pool. */ int zpool_scan(zpool_handle_t *zhp, pool_scan_func_t func, pool_scrub_cmd_t cmd) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; int err; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = func; zc.zc_flags = cmd; if (zfs_ioctl(hdl, ZFS_IOC_POOL_SCAN, &zc) == 0) return (0); err = errno; /* ECANCELED on a scrub means we resumed a paused scrub */ if (err == ECANCELED && func == POOL_SCAN_SCRUB && cmd == POOL_SCRUB_NORMAL) return (0); if (err == ENOENT && func != POOL_SCAN_NONE && cmd == POOL_SCRUB_NORMAL) return (0); if (func == POOL_SCAN_SCRUB) { if (cmd == POOL_SCRUB_PAUSE) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot pause scrubbing %s"), zc.zc_name); } else { assert(cmd == POOL_SCRUB_NORMAL); (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot scrub %s"), zc.zc_name); } } else if (func == POOL_SCAN_RESILVER) { assert(cmd == POOL_SCRUB_NORMAL); (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot restart resilver on %s"), zc.zc_name); } else if (func == POOL_SCAN_NONE) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot cancel scrubbing %s"), zc.zc_name); } else { assert(!"unexpected result"); } if (err == EBUSY) { nvlist_t *nvroot; pool_scan_stat_t *ps = NULL; uint_t psc; nvroot = fnvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE); (void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS, (uint64_t **)&ps, &psc); if (ps && ps->pss_func == POOL_SCAN_SCRUB && ps->pss_state == DSS_SCANNING) { if (cmd == POOL_SCRUB_PAUSE) return (zfs_error(hdl, EZFS_SCRUB_PAUSED, errbuf)); else return (zfs_error(hdl, EZFS_SCRUBBING, errbuf)); } else { return (zfs_error(hdl, EZFS_RESILVERING, errbuf)); } } else if (err == ENOENT) { return (zfs_error(hdl, EZFS_NO_SCRUB, errbuf)); } else if (err == ENOTSUP && func == POOL_SCAN_RESILVER) { return (zfs_error(hdl, EZFS_NO_RESILVER_DEFER, errbuf)); } else { return (zpool_standard_error(hdl, err, errbuf)); } } /* * Find a vdev that matches the search criteria specified. We use the * the nvpair name to determine how we should look for the device. * 'avail_spare' is set to TRUE if the provided guid refers to an AVAIL * spare; but FALSE if its an INUSE spare. */ static nvlist_t * vdev_to_nvlist_iter(nvlist_t *nv, nvlist_t *search, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { uint_t c, children; nvlist_t **child; nvlist_t *ret; uint64_t is_log; char *srchkey; nvpair_t *pair = nvlist_next_nvpair(search, NULL); /* Nothing to look for */ if (search == NULL || pair == NULL) return (NULL); /* Obtain the key we will use to search */ srchkey = nvpair_name(pair); switch (nvpair_type(pair)) { case DATA_TYPE_UINT64: if (strcmp(srchkey, ZPOOL_CONFIG_GUID) == 0) { uint64_t srchval = fnvpair_value_uint64(pair); uint64_t theguid = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID); if (theguid == srchval) return (nv); } break; case DATA_TYPE_STRING: { char *srchval, *val; srchval = fnvpair_value_string(pair); if (nvlist_lookup_string(nv, srchkey, &val) != 0) break; /* * Search for the requested value. Special cases: * * - ZPOOL_CONFIG_PATH for whole disk entries. These end in * "-part1", or "p1". The suffix is hidden from the user, * but included in the string, so this matches around it. * - ZPOOL_CONFIG_PATH for short names zfs_strcmp_shortname() * is used to check all possible expanded paths. * - looking for a top-level vdev name (i.e. ZPOOL_CONFIG_TYPE). * * Otherwise, all other searches are simple string compares. */ if (strcmp(srchkey, ZPOOL_CONFIG_PATH) == 0) { uint64_t wholedisk = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk); if (zfs_strcmp_pathname(srchval, val, wholedisk) == 0) return (nv); } else if (strcmp(srchkey, ZPOOL_CONFIG_TYPE) == 0 && val) { char *type, *idx, *end, *p; uint64_t id, vdev_id; /* * Determine our vdev type, keeping in mind * that the srchval is composed of a type and * vdev id pair (i.e. mirror-4). */ if ((type = strdup(srchval)) == NULL) return (NULL); if ((p = strrchr(type, '-')) == NULL) { free(type); break; } idx = p + 1; *p = '\0'; /* * If the types don't match then keep looking. */ if (strncmp(val, type, strlen(val)) != 0) { free(type); break; } verify(zpool_vdev_is_interior(type)); id = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID); errno = 0; vdev_id = strtoull(idx, &end, 10); /* * If we are looking for a raidz and a parity is * specified, make sure it matches. */ int rzlen = strlen(VDEV_TYPE_RAIDZ); assert(rzlen == strlen(VDEV_TYPE_DRAID)); int typlen = strlen(type); if ((strncmp(type, VDEV_TYPE_RAIDZ, rzlen) == 0 || strncmp(type, VDEV_TYPE_DRAID, rzlen) == 0) && typlen != rzlen) { uint64_t vdev_parity; int parity = *(type + rzlen) - '0'; if (parity <= 0 || parity > 3 || (typlen - rzlen) != 1) { /* * Nonsense parity specified, can * never match */ free(type); return (NULL); } vdev_parity = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY); if ((int)vdev_parity != parity) { free(type); break; } } free(type); if (errno != 0) return (NULL); /* * Now verify that we have the correct vdev id. */ if (vdev_id == id) return (nv); } /* * Common case */ if (strcmp(srchval, val) == 0) return (nv); break; } default: break; } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (NULL); for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { /* * The 'is_log' value is only set for the toplevel * vdev, not the leaf vdevs. So we always lookup the * log device from the root of the vdev tree (where * 'log' is non-NULL). */ if (log != NULL && nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, &is_log) == 0 && is_log) { *log = B_TRUE; } return (ret); } } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, &child, &children) == 0) { for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { *avail_spare = B_TRUE; return (ret); } } } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, &child, &children) == 0) { for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { *l2cache = B_TRUE; return (ret); } } } return (NULL); } /* * Given a physical path or guid, find the associated vdev. */ nvlist_t * zpool_find_vdev_by_physpath(zpool_handle_t *zhp, const char *ppath, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { nvlist_t *search, *nvroot, *ret; uint64_t guid; char *end; search = fnvlist_alloc(); guid = strtoull(ppath, &end, 0); if (guid != 0 && *end == '\0') { fnvlist_add_uint64(search, ZPOOL_CONFIG_GUID, guid); } else { fnvlist_add_string(search, ZPOOL_CONFIG_PHYS_PATH, ppath); } nvroot = fnvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE); *avail_spare = B_FALSE; *l2cache = B_FALSE; if (log != NULL) *log = B_FALSE; ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log); fnvlist_free(search); return (ret); } /* * Determine if we have an "interior" top-level vdev (i.e mirror/raidz). */ static boolean_t zpool_vdev_is_interior(const char *name) { if (strncmp(name, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0 || strncmp(name, VDEV_TYPE_SPARE, strlen(VDEV_TYPE_SPARE)) == 0 || strncmp(name, VDEV_TYPE_REPLACING, strlen(VDEV_TYPE_REPLACING)) == 0 || strncmp(name, VDEV_TYPE_MIRROR, strlen(VDEV_TYPE_MIRROR)) == 0) return (B_TRUE); if (strncmp(name, VDEV_TYPE_DRAID, strlen(VDEV_TYPE_DRAID)) == 0 && !zpool_is_draid_spare(name)) return (B_TRUE); return (B_FALSE); } nvlist_t * zpool_find_vdev(zpool_handle_t *zhp, const char *path, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { char *end; nvlist_t *nvroot, *search, *ret; uint64_t guid; search = fnvlist_alloc(); guid = strtoull(path, &end, 0); if (guid != 0 && *end == '\0') { fnvlist_add_uint64(search, ZPOOL_CONFIG_GUID, guid); } else if (zpool_vdev_is_interior(path)) { fnvlist_add_string(search, ZPOOL_CONFIG_TYPE, path); } else { fnvlist_add_string(search, ZPOOL_CONFIG_PATH, path); } nvroot = fnvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE); *avail_spare = B_FALSE; *l2cache = B_FALSE; if (log != NULL) *log = B_FALSE; ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log); fnvlist_free(search); return (ret); } /* * Convert a vdev path to a GUID. Returns GUID or 0 on error. * * If is_spare, is_l2cache, or is_log is non-NULL, then store within it * if the VDEV is a spare, l2cache, or log device. If they're NULL then * ignore them. */ static uint64_t zpool_vdev_path_to_guid_impl(zpool_handle_t *zhp, const char *path, boolean_t *is_spare, boolean_t *is_l2cache, boolean_t *is_log) { boolean_t spare = B_FALSE, l2cache = B_FALSE, log = B_FALSE; nvlist_t *tgt; if ((tgt = zpool_find_vdev(zhp, path, &spare, &l2cache, &log)) == NULL) return (0); if (is_spare != NULL) *is_spare = spare; if (is_l2cache != NULL) *is_l2cache = l2cache; if (is_log != NULL) *is_log = log; return (fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID)); } /* Convert a vdev path to a GUID. Returns GUID or 0 on error. */ uint64_t zpool_vdev_path_to_guid(zpool_handle_t *zhp, const char *path) { return (zpool_vdev_path_to_guid_impl(zhp, path, NULL, NULL, NULL)); } /* * Bring the specified vdev online. The 'flags' parameter is a set of the * ZFS_ONLINE_* flags. */ int zpool_vdev_online(zpool_handle_t *zhp, const char *path, int flags, vdev_state_t *newstate) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; libzfs_handle_t *hdl = zhp->zpool_hdl; if (flags & ZFS_ONLINE_EXPAND) { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot expand %s"), path); } else { (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot online %s"), path); } (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, &islog)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); #ifndef __FreeBSD__ char *pathname; if ((flags & ZFS_ONLINE_EXPAND || zpool_get_prop_int(zhp, ZPOOL_PROP_AUTOEXPAND, NULL)) && nvlist_lookup_string(tgt, ZPOOL_CONFIG_PATH, &pathname) == 0) { uint64_t wholedisk = 0; (void) nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk); /* * XXX - L2ARC 1.0 devices can't support expansion. */ if (l2cache) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot expand cache devices")); return (zfs_error(hdl, EZFS_VDEVNOTSUP, errbuf)); } if (wholedisk) { const char *fullpath = path; char buf[MAXPATHLEN]; int error; if (path[0] != '/') { error = zfs_resolve_shortname(path, buf, sizeof (buf)); if (error != 0) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); fullpath = buf; } error = zpool_relabel_disk(hdl, fullpath, errbuf); if (error != 0) return (error); } } #endif zc.zc_cookie = VDEV_STATE_ONLINE; zc.zc_obj = flags; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) != 0) { if (errno == EINVAL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "was split " "from this pool into a new one. Use '%s' " "instead"), "zpool detach"); return (zfs_error(hdl, EZFS_POSTSPLIT_ONLINE, errbuf)); } return (zpool_standard_error(hdl, errno, errbuf)); } *newstate = zc.zc_cookie; return (0); } /* * Take the specified vdev offline */ int zpool_vdev_offline(zpool_handle_t *zhp, const char *path, boolean_t istmp) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; nvlist_t *tgt; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot offline %s"), path); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); zc.zc_cookie = VDEV_STATE_OFFLINE; zc.zc_obj = istmp ? ZFS_OFFLINE_TEMPORARY : 0; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); switch (errno) { case EBUSY: /* * There are no other replicas of this device. */ return (zfs_error(hdl, EZFS_NOREPLICAS, errbuf)); case EEXIST: /* * The log device has unplayed logs */ return (zfs_error(hdl, EZFS_UNPLAYED_LOGS, errbuf)); default: return (zpool_standard_error(hdl, errno, errbuf)); } } +/* + * Remove the specified vdev asynchronously from the configuration, so + * that it may come ONLINE if reinserted. This is called from zed on + * Udev remove event. + * Note: We also have a similar function zpool_vdev_remove() that + * removes the vdev from the pool. + */ +int +zpool_vdev_remove_wanted(zpool_handle_t *zhp, const char *path) +{ + zfs_cmd_t zc = {"\0"}; + char errbuf[ERRBUFLEN]; + nvlist_t *tgt; + boolean_t avail_spare, l2cache; + libzfs_handle_t *hdl = zhp->zpool_hdl; + + (void) snprintf(errbuf, sizeof (errbuf), + dgettext(TEXT_DOMAIN, "cannot remove %s"), path); + + (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); + if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, + NULL)) == NULL) + return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); + + zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); + + if (avail_spare) + return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); + + zc.zc_cookie = VDEV_STATE_REMOVED; + + if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) + return (0); + + return (zpool_standard_error(hdl, errno, errbuf)); +} + /* * Mark the given vdev faulted. */ int zpool_vdev_fault(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot fault %llu"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = VDEV_STATE_FAULTED; zc.zc_obj = aux; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); switch (errno) { case EBUSY: /* * There are no other replicas of this device. */ return (zfs_error(hdl, EZFS_NOREPLICAS, errbuf)); default: return (zpool_standard_error(hdl, errno, errbuf)); } } /* * Mark the given vdev degraded. */ int zpool_vdev_degrade(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot degrade %llu"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = VDEV_STATE_DEGRADED; zc.zc_obj = aux; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, errbuf)); } /* * Returns TRUE if the given nvlist is a vdev that was originally swapped in as * a hot spare. */ static boolean_t is_replacing_spare(nvlist_t *search, nvlist_t *tgt, int which) { nvlist_t **child; uint_t c, children; if (nvlist_lookup_nvlist_array(search, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { char *type = fnvlist_lookup_string(search, ZPOOL_CONFIG_TYPE); if ((strcmp(type, VDEV_TYPE_SPARE) == 0 || strcmp(type, VDEV_TYPE_DRAID_SPARE) == 0) && children == 2 && child[which] == tgt) return (B_TRUE); for (c = 0; c < children; c++) if (is_replacing_spare(child[c], tgt, which)) return (B_TRUE); } return (B_FALSE); } /* * Attach new_disk (fully described by nvroot) to old_disk. * If 'replacing' is specified, the new disk will replace the old one. */ int zpool_vdev_attach(zpool_handle_t *zhp, const char *old_disk, const char *new_disk, nvlist_t *nvroot, int replacing, boolean_t rebuild) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; int ret; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; uint64_t val; char *newname; nvlist_t **child; uint_t children; nvlist_t *config_root; libzfs_handle_t *hdl = zhp->zpool_hdl; if (replacing) (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot replace %s with %s"), old_disk, new_disk); else (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot attach %s to %s"), new_disk, old_disk); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, old_disk, &avail_spare, &l2cache, &islog)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); if (l2cache) return (zfs_error(hdl, EZFS_ISL2CACHE, errbuf)); zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); zc.zc_cookie = replacing; zc.zc_simple = rebuild; if (rebuild && zfeature_lookup_guid("org.openzfs:device_rebuild", NULL) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs module doesn't support device rebuilds")); return (zfs_error(hdl, EZFS_POOL_NOTSUP, errbuf)); } if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0 || children != 1) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device must be a single disk")); return (zfs_error(hdl, EZFS_INVALCONFIG, errbuf)); } config_root = fnvlist_lookup_nvlist(zpool_get_config(zhp, NULL), ZPOOL_CONFIG_VDEV_TREE); if ((newname = zpool_vdev_name(NULL, NULL, child[0], 0)) == NULL) return (-1); /* * If the target is a hot spare that has been swapped in, we can only * replace it with another hot spare. */ if (replacing && nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_IS_SPARE, &val) == 0 && (zpool_find_vdev(zhp, newname, &avail_spare, &l2cache, NULL) == NULL || !avail_spare) && is_replacing_spare(config_root, tgt, 1)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "can only be replaced by another hot spare")); free(newname); return (zfs_error(hdl, EZFS_BADTARGET, errbuf)); } free(newname); zcmd_write_conf_nvlist(hdl, &zc, nvroot); ret = zfs_ioctl(hdl, ZFS_IOC_VDEV_ATTACH, &zc); zcmd_free_nvlists(&zc); if (ret == 0) return (0); switch (errno) { case ENOTSUP: /* * Can't attach to or replace this type of vdev. */ if (replacing) { uint64_t version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if (islog) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot replace a log with a spare")); } else if (rebuild) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "only mirror and dRAID vdevs support " "sequential reconstruction")); } else if (zpool_is_draid_spare(new_disk)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dRAID spares can only replace child " "devices in their parent's dRAID vdev")); } else if (version >= SPA_VERSION_MULTI_REPLACE) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "already in replacing/spare config; wait " "for completion or use 'zpool detach'")); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot replace a replacing device")); } } else { char status[64] = {0}; zpool_prop_get_feature(zhp, "feature@device_rebuild", status, 63); if (rebuild && strncmp(status, ZFS_FEATURE_DISABLED, 64) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device_rebuild feature must be enabled " "in order to use sequential " "reconstruction")); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "can only attach to mirrors and top-level " "disks")); } } (void) zfs_error(hdl, EZFS_BADTARGET, errbuf); break; case EINVAL: /* * The new device must be a single disk. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device must be a single disk")); (void) zfs_error(hdl, EZFS_INVALCONFIG, errbuf); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s is busy, " "or device removal is in progress"), new_disk); (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case EOVERFLOW: /* * The new device is too small. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device is too small")); (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case EDOM: /* * The new device has a different optimal sector size. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device has a different optimal sector size; use the " "option '-o ashift=N' to override the optimal size")); (void) zfs_error(hdl, EZFS_BADDEV, errbuf); break; case ENAMETOOLONG: /* * The resulting top-level vdev spec won't fit in the label. */ (void) zfs_error(hdl, EZFS_DEVOVERFLOW, errbuf); break; default: (void) zpool_standard_error(hdl, errno, errbuf); } return (-1); } /* * Detach the specified device. */ int zpool_vdev_detach(zpool_handle_t *zhp, const char *path) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; nvlist_t *tgt; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot detach %s"), path); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); if (l2cache) return (zfs_error(hdl, EZFS_ISL2CACHE, errbuf)); zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_DETACH, &zc) == 0) return (0); switch (errno) { case ENOTSUP: /* * Can't detach from this type of vdev. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "only " "applicable to mirror and replacing vdevs")); (void) zfs_error(hdl, EZFS_BADTARGET, errbuf); break; case EBUSY: /* * There are no other replicas of this device. */ (void) zfs_error(hdl, EZFS_NOREPLICAS, errbuf); break; default: (void) zpool_standard_error(hdl, errno, errbuf); } return (-1); } /* * Find a mirror vdev in the source nvlist. * * The mchild array contains a list of disks in one of the top-level mirrors * of the source pool. The schild array contains a list of disks that the * user specified on the command line. We loop over the mchild array to * see if any entry in the schild array matches. * * If a disk in the mchild array is found in the schild array, we return * the index of that entry. Otherwise we return -1. */ static int find_vdev_entry(zpool_handle_t *zhp, nvlist_t **mchild, uint_t mchildren, nvlist_t **schild, uint_t schildren) { uint_t mc; for (mc = 0; mc < mchildren; mc++) { uint_t sc; char *mpath = zpool_vdev_name(zhp->zpool_hdl, zhp, mchild[mc], 0); for (sc = 0; sc < schildren; sc++) { char *spath = zpool_vdev_name(zhp->zpool_hdl, zhp, schild[sc], 0); boolean_t result = (strcmp(mpath, spath) == 0); free(spath); if (result) { free(mpath); return (mc); } } free(mpath); } return (-1); } /* * Split a mirror pool. If newroot points to null, then a new nvlist * is generated and it is the responsibility of the caller to free it. */ int zpool_vdev_split(zpool_handle_t *zhp, char *newname, nvlist_t **newroot, nvlist_t *props, splitflags_t flags) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN], *bias; nvlist_t *tree, *config, **child, **newchild, *newconfig = NULL; nvlist_t **varray = NULL, *zc_props = NULL; uint_t c, children, newchildren, lastlog = 0, vcount, found = 0; libzfs_handle_t *hdl = zhp->zpool_hdl; uint64_t vers, readonly = B_FALSE; boolean_t freelist = B_FALSE, memory_err = B_TRUE; int retval = 0; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "Unable to split %s"), zhp->zpool_name); if (!zpool_name_valid(hdl, B_FALSE, newname)) return (zfs_error(hdl, EZFS_INVALIDNAME, errbuf)); if ((config = zpool_get_config(zhp, NULL)) == NULL) { (void) fprintf(stderr, gettext("Internal error: unable to " "retrieve pool configuration\n")); return (-1); } tree = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE); vers = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION); if (props) { prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE }; if ((zc_props = zpool_valid_proplist(hdl, zhp->zpool_name, props, vers, flags, errbuf)) == NULL) return (-1); (void) nvlist_lookup_uint64(zc_props, zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly); if (readonly) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property %s can only be set at import time"), zpool_prop_to_name(ZPOOL_PROP_READONLY)); return (-1); } } if (nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Source pool is missing vdev tree")); nvlist_free(zc_props); return (-1); } varray = zfs_alloc(hdl, children * sizeof (nvlist_t *)); vcount = 0; if (*newroot == NULL || nvlist_lookup_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN, &newchild, &newchildren) != 0) newchildren = 0; for (c = 0; c < children; c++) { uint64_t is_log = B_FALSE, is_hole = B_FALSE; boolean_t is_special = B_FALSE, is_dedup = B_FALSE; char *type; nvlist_t **mchild, *vdev; uint_t mchildren; int entry; /* * Unlike cache & spares, slogs are stored in the * ZPOOL_CONFIG_CHILDREN array. We filter them out here. */ (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, &is_log); (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE, &is_hole); if (is_log || is_hole) { /* * Create a hole vdev and put it in the config. */ if (nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) != 0) goto out; if (nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE, VDEV_TYPE_HOLE) != 0) goto out; if (nvlist_add_uint64(vdev, ZPOOL_CONFIG_IS_HOLE, 1) != 0) goto out; if (lastlog == 0) lastlog = vcount; varray[vcount++] = vdev; continue; } lastlog = 0; type = fnvlist_lookup_string(child[c], ZPOOL_CONFIG_TYPE); if (strcmp(type, VDEV_TYPE_INDIRECT) == 0) { vdev = child[c]; if (nvlist_dup(vdev, &varray[vcount++], 0) != 0) goto out; continue; } else if (strcmp(type, VDEV_TYPE_MIRROR) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Source pool must be composed only of mirrors\n")); retval = zfs_error(hdl, EZFS_INVALCONFIG, errbuf); goto out; } if (nvlist_lookup_string(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0) { if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) is_special = B_TRUE; else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0) is_dedup = B_TRUE; } verify(nvlist_lookup_nvlist_array(child[c], ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren) == 0); /* find or add an entry for this top-level vdev */ if (newchildren > 0 && (entry = find_vdev_entry(zhp, mchild, mchildren, newchild, newchildren)) >= 0) { /* We found a disk that the user specified. */ vdev = mchild[entry]; ++found; } else { /* User didn't specify a disk for this vdev. */ vdev = mchild[mchildren - 1]; } if (nvlist_dup(vdev, &varray[vcount++], 0) != 0) goto out; if (flags.dryrun != 0) { if (is_dedup == B_TRUE) { if (nvlist_add_string(varray[vcount - 1], ZPOOL_CONFIG_ALLOCATION_BIAS, VDEV_ALLOC_BIAS_DEDUP) != 0) goto out; } else if (is_special == B_TRUE) { if (nvlist_add_string(varray[vcount - 1], ZPOOL_CONFIG_ALLOCATION_BIAS, VDEV_ALLOC_BIAS_SPECIAL) != 0) goto out; } } } /* did we find every disk the user specified? */ if (found != newchildren) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Device list must " "include at most one disk from each mirror")); retval = zfs_error(hdl, EZFS_INVALCONFIG, errbuf); goto out; } /* Prepare the nvlist for populating. */ if (*newroot == NULL) { if (nvlist_alloc(newroot, NV_UNIQUE_NAME, 0) != 0) goto out; freelist = B_TRUE; if (nvlist_add_string(*newroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) != 0) goto out; } else { verify(nvlist_remove_all(*newroot, ZPOOL_CONFIG_CHILDREN) == 0); } /* Add all the children we found */ if (nvlist_add_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)varray, lastlog == 0 ? vcount : lastlog) != 0) goto out; /* * If we're just doing a dry run, exit now with success. */ if (flags.dryrun) { memory_err = B_FALSE; freelist = B_FALSE; goto out; } /* now build up the config list & call the ioctl */ if (nvlist_alloc(&newconfig, NV_UNIQUE_NAME, 0) != 0) goto out; if (nvlist_add_nvlist(newconfig, ZPOOL_CONFIG_VDEV_TREE, *newroot) != 0 || nvlist_add_string(newconfig, ZPOOL_CONFIG_POOL_NAME, newname) != 0 || nvlist_add_uint64(newconfig, ZPOOL_CONFIG_VERSION, vers) != 0) goto out; /* * The new pool is automatically part of the namespace unless we * explicitly export it. */ if (!flags.import) zc.zc_cookie = ZPOOL_EXPORT_AFTER_SPLIT; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); (void) strlcpy(zc.zc_string, newname, sizeof (zc.zc_string)); zcmd_write_conf_nvlist(hdl, &zc, newconfig); if (zc_props != NULL) zcmd_write_src_nvlist(hdl, &zc, zc_props); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SPLIT, &zc) != 0) { retval = zpool_standard_error(hdl, errno, errbuf); goto out; } freelist = B_FALSE; memory_err = B_FALSE; out: if (varray != NULL) { int v; for (v = 0; v < vcount; v++) nvlist_free(varray[v]); free(varray); } zcmd_free_nvlists(&zc); nvlist_free(zc_props); nvlist_free(newconfig); if (freelist) { nvlist_free(*newroot); *newroot = NULL; } if (retval != 0) return (retval); if (memory_err) return (no_memory(hdl)); return (0); } /* * Remove the given device. */ int zpool_vdev_remove(zpool_handle_t *zhp, const char *path) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; libzfs_handle_t *hdl = zhp->zpool_hdl; uint64_t version; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot remove %s"), path); if (zpool_is_draid_spare(path)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "dRAID spares cannot be removed")); return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); } (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, &islog)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if (islog && version < SPA_VERSION_HOLES) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgraded to support log removal")); return (zfs_error(hdl, EZFS_BADVERSION, errbuf)); } zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_REMOVE, &zc) == 0) return (0); switch (errno) { case EINVAL: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid config; all top-level vdevs must " "have the same sector size and not be raidz.")); (void) zfs_error(hdl, EZFS_INVALCONFIG, errbuf); break; case EBUSY: if (islog) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Mount encrypted datasets to replay logs.")); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Pool busy; removal may already be in progress")); } (void) zfs_error(hdl, EZFS_BUSY, errbuf); break; case EACCES: if (islog) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Mount encrypted datasets to replay logs.")); (void) zfs_error(hdl, EZFS_BUSY, errbuf); } else { (void) zpool_standard_error(hdl, errno, errbuf); } break; default: (void) zpool_standard_error(hdl, errno, errbuf); } return (-1); } int zpool_vdev_remove_cancel(zpool_handle_t *zhp) { zfs_cmd_t zc = {{0}}; char errbuf[ERRBUFLEN]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot cancel removal")); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = 1; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_REMOVE, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, errbuf)); } int zpool_vdev_indirect_size(zpool_handle_t *zhp, const char *path, uint64_t *sizep) { char errbuf[ERRBUFLEN]; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot determine indirect size of %s"), path); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, &islog)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); if (avail_spare || l2cache || islog) { *sizep = 0; return (0); } if (nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_INDIRECT_SIZE, sizep) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "indirect size not available")); return (zfs_error(hdl, EINVAL, errbuf)); } return (0); } /* * Clear the errors for the pool, or the particular device if specified. */ int zpool_clear(zpool_handle_t *zhp, const char *path, nvlist_t *rewindnvl) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; nvlist_t *tgt; zpool_load_policy_t policy; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; nvlist_t *nvi = NULL; int error; if (path) (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot clear errors for %s"), path); else (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot clear errors for %s"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (path) { if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, errbuf)); /* * Don't allow error clearing for hot spares. Do allow * error clearing for l2cache devices. */ if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, errbuf)); zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); } zpool_get_load_policy(rewindnvl, &policy); zc.zc_cookie = policy.zlp_rewind; zcmd_alloc_dst_nvlist(hdl, &zc, zhp->zpool_config_size * 2); zcmd_write_src_nvlist(hdl, &zc, rewindnvl); while ((error = zfs_ioctl(hdl, ZFS_IOC_CLEAR, &zc)) != 0 && errno == ENOMEM) zcmd_expand_dst_nvlist(hdl, &zc); if (!error || ((policy.zlp_rewind & ZPOOL_TRY_REWIND) && errno != EPERM && errno != EACCES)) { if (policy.zlp_rewind & (ZPOOL_DO_REWIND | ZPOOL_TRY_REWIND)) { (void) zcmd_read_dst_nvlist(hdl, &zc, &nvi); zpool_rewind_exclaim(hdl, zc.zc_name, ((policy.zlp_rewind & ZPOOL_TRY_REWIND) != 0), nvi); nvlist_free(nvi); } zcmd_free_nvlists(&zc); return (0); } zcmd_free_nvlists(&zc); return (zpool_standard_error(hdl, errno, errbuf)); } /* * Similar to zpool_clear(), but takes a GUID (used by fmd). */ int zpool_vdev_clear(zpool_handle_t *zhp, uint64_t guid) { zfs_cmd_t zc = {"\0"}; char errbuf[ERRBUFLEN]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot clear errors for %llx"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = ZPOOL_NO_REWIND; if (zfs_ioctl(hdl, ZFS_IOC_CLEAR, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, errbuf)); } /* * Change the GUID for a pool. */ int zpool_reguid(zpool_handle_t *zhp) { char errbuf[ERRBUFLEN]; libzfs_handle_t *hdl = zhp->zpool_hdl; zfs_cmd_t zc = {"\0"}; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot reguid '%s'"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zfs_ioctl(hdl, ZFS_IOC_POOL_REGUID, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, errbuf)); } /* * Reopen the pool. */ int zpool_reopen_one(zpool_handle_t *zhp, void *data) { libzfs_handle_t *hdl = zpool_get_handle(zhp); const char *pool_name = zpool_get_name(zhp); boolean_t *scrub_restart = data; int error; error = lzc_reopen(pool_name, *scrub_restart); if (error) { return (zpool_standard_error_fmt(hdl, error, dgettext(TEXT_DOMAIN, "cannot reopen '%s'"), pool_name)); } return (0); } /* call into libzfs_core to execute the sync IOCTL per pool */ int zpool_sync_one(zpool_handle_t *zhp, void *data) { int ret; libzfs_handle_t *hdl = zpool_get_handle(zhp); const char *pool_name = zpool_get_name(zhp); boolean_t *force = data; nvlist_t *innvl = fnvlist_alloc(); fnvlist_add_boolean_value(innvl, "force", *force); if ((ret = lzc_sync(pool_name, innvl, NULL)) != 0) { nvlist_free(innvl); return (zpool_standard_error_fmt(hdl, ret, dgettext(TEXT_DOMAIN, "sync '%s' failed"), pool_name)); } nvlist_free(innvl); return (0); } #define PATH_BUF_LEN 64 /* * Given a vdev, return the name to display in iostat. If the vdev has a path, * we use that, stripping off any leading "/dev/dsk/"; if not, we use the type. * We also check if this is a whole disk, in which case we strip off the * trailing 's0' slice name. * * This routine is also responsible for identifying when disks have been * reconfigured in a new location. The kernel will have opened the device by * devid, but the path will still refer to the old location. To catch this, we * first do a path -> devid translation (which is fast for the common case). If * the devid matches, we're done. If not, we do a reverse devid -> path * translation and issue the appropriate ioctl() to update the path of the vdev. * If 'zhp' is NULL, then this is an exported pool, and we don't need to do any * of these checks. */ char * zpool_vdev_name(libzfs_handle_t *hdl, zpool_handle_t *zhp, nvlist_t *nv, int name_flags) { char *type, *tpath; const char *path; uint64_t value; char buf[PATH_BUF_LEN]; char tmpbuf[PATH_BUF_LEN * 2]; /* * vdev_name will be "root"/"root-0" for the root vdev, but it is the * zpool name that will be displayed to the user. */ type = fnvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE); if (zhp != NULL && strcmp(type, "root") == 0) return (zfs_strdup(hdl, zpool_get_name(zhp))); if (libzfs_envvar_is_set("ZPOOL_VDEV_NAME_PATH")) name_flags |= VDEV_NAME_PATH; if (libzfs_envvar_is_set("ZPOOL_VDEV_NAME_GUID")) name_flags |= VDEV_NAME_GUID; if (libzfs_envvar_is_set("ZPOOL_VDEV_NAME_FOLLOW_LINKS")) name_flags |= VDEV_NAME_FOLLOW_LINKS; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &value) == 0 || name_flags & VDEV_NAME_GUID) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value); (void) snprintf(buf, sizeof (buf), "%llu", (u_longlong_t)value); path = buf; } else if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &tpath) == 0) { path = tpath; if (name_flags & VDEV_NAME_FOLLOW_LINKS) { char *rp = realpath(path, NULL); if (rp) { strlcpy(buf, rp, sizeof (buf)); path = buf; free(rp); } } /* * For a block device only use the name. */ if ((strcmp(type, VDEV_TYPE_DISK) == 0) && !(name_flags & VDEV_NAME_PATH)) { path = zfs_strip_path(path); } /* * Remove the partition from the path if this is a whole disk. */ if (strcmp(type, VDEV_TYPE_DRAID_SPARE) != 0 && nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &value) == 0 && value && !(name_flags & VDEV_NAME_PATH)) { return (zfs_strip_partition(path)); } } else { path = type; /* * If it's a raidz device, we need to stick in the parity level. */ if (strcmp(path, VDEV_TYPE_RAIDZ) == 0) { value = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY); (void) snprintf(buf, sizeof (buf), "%s%llu", path, (u_longlong_t)value); path = buf; } /* * If it's a dRAID device, we add parity, groups, and spares. */ if (strcmp(path, VDEV_TYPE_DRAID) == 0) { uint64_t ndata, nparity, nspares; nvlist_t **child; uint_t children; verify(nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0); nparity = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY); ndata = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA); nspares = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NSPARES); path = zpool_draid_name(buf, sizeof (buf), ndata, nparity, nspares, children); } /* * We identify each top-level vdev by using a * naming convention. */ if (name_flags & VDEV_NAME_TYPE_ID) { uint64_t id = fnvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID); (void) snprintf(tmpbuf, sizeof (tmpbuf), "%s-%llu", path, (u_longlong_t)id); path = tmpbuf; } } return (zfs_strdup(hdl, path)); } static int zbookmark_mem_compare(const void *a, const void *b) { return (memcmp(a, b, sizeof (zbookmark_phys_t))); } /* * Retrieve the persistent error log, uniquify the members, and return to the * caller. */ int zpool_get_errlog(zpool_handle_t *zhp, nvlist_t **nverrlistp) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; uint64_t count; zbookmark_phys_t *zb = NULL; int i; /* * Retrieve the raw error list from the kernel. If the number of errors * has increased, allocate more space and continue until we get the * entire list. */ count = fnvlist_lookup_uint64(zhp->zpool_config, ZPOOL_CONFIG_ERRCOUNT); if (count == 0) return (0); zc.zc_nvlist_dst = (uintptr_t)zfs_alloc(zhp->zpool_hdl, count * sizeof (zbookmark_phys_t)); zc.zc_nvlist_dst_size = count; (void) strcpy(zc.zc_name, zhp->zpool_name); for (;;) { if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_ERROR_LOG, &zc) != 0) { free((void *)(uintptr_t)zc.zc_nvlist_dst); if (errno == ENOMEM) { void *dst; count = zc.zc_nvlist_dst_size; dst = zfs_alloc(zhp->zpool_hdl, count * sizeof (zbookmark_phys_t)); zc.zc_nvlist_dst = (uintptr_t)dst; } else { return (zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "errors: List of " "errors unavailable"))); } } else { break; } } /* * Sort the resulting bookmarks. This is a little confusing due to the * implementation of ZFS_IOC_ERROR_LOG. The bookmarks are copied last * to first, and 'zc_nvlist_dst_size' indicates the number of bookmarks * _not_ copied as part of the process. So we point the start of our * array appropriate and decrement the total number of elements. */ zb = ((zbookmark_phys_t *)(uintptr_t)zc.zc_nvlist_dst) + zc.zc_nvlist_dst_size; count -= zc.zc_nvlist_dst_size; qsort(zb, count, sizeof (zbookmark_phys_t), zbookmark_mem_compare); verify(nvlist_alloc(nverrlistp, 0, KM_SLEEP) == 0); /* * Fill in the nverrlistp with nvlist's of dataset and object numbers. */ for (i = 0; i < count; i++) { nvlist_t *nv; /* ignoring zb_blkid and zb_level for now */ if (i > 0 && zb[i-1].zb_objset == zb[i].zb_objset && zb[i-1].zb_object == zb[i].zb_object) continue; if (nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) != 0) goto nomem; if (nvlist_add_uint64(nv, ZPOOL_ERR_DATASET, zb[i].zb_objset) != 0) { nvlist_free(nv); goto nomem; } if (nvlist_add_uint64(nv, ZPOOL_ERR_OBJECT, zb[i].zb_object) != 0) { nvlist_free(nv); goto nomem; } if (nvlist_add_nvlist(*nverrlistp, "ejk", nv) != 0) { nvlist_free(nv); goto nomem; } nvlist_free(nv); } free((void *)(uintptr_t)zc.zc_nvlist_dst); return (0); nomem: free((void *)(uintptr_t)zc.zc_nvlist_dst); return (no_memory(zhp->zpool_hdl)); } /* * Upgrade a ZFS pool to the latest on-disk version. */ int zpool_upgrade(zpool_handle_t *zhp, uint64_t new_version) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strcpy(zc.zc_name, zhp->zpool_name); zc.zc_cookie = new_version; if (zfs_ioctl(hdl, ZFS_IOC_POOL_UPGRADE, &zc) != 0) return (zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot upgrade '%s'"), zhp->zpool_name)); return (0); } void zfs_save_arguments(int argc, char **argv, char *string, int len) { int i; (void) strlcpy(string, zfs_basename(argv[0]), len); for (i = 1; i < argc; i++) { (void) strlcat(string, " ", len); (void) strlcat(string, argv[i], len); } } int zpool_log_history(libzfs_handle_t *hdl, const char *message) { zfs_cmd_t zc = {"\0"}; nvlist_t *args; args = fnvlist_alloc(); fnvlist_add_string(args, "message", message); zcmd_write_src_nvlist(hdl, &zc, args); int err = zfs_ioctl(hdl, ZFS_IOC_LOG_HISTORY, &zc); nvlist_free(args); zcmd_free_nvlists(&zc); return (err); } /* * Perform ioctl to get some command history of a pool. * * 'buf' is the buffer to fill up to 'len' bytes. 'off' is the * logical offset of the history buffer to start reading from. * * Upon return, 'off' is the next logical offset to read from and * 'len' is the actual amount of bytes read into 'buf'. */ static int get_history(zpool_handle_t *zhp, char *buf, uint64_t *off, uint64_t *len) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_history = (uint64_t)(uintptr_t)buf; zc.zc_history_len = *len; zc.zc_history_offset = *off; if (zfs_ioctl(hdl, ZFS_IOC_POOL_GET_HISTORY, &zc) != 0) { switch (errno) { case EPERM: return (zfs_error_fmt(hdl, EZFS_PERM, dgettext(TEXT_DOMAIN, "cannot show history for pool '%s'"), zhp->zpool_name)); case ENOENT: return (zfs_error_fmt(hdl, EZFS_NOHISTORY, dgettext(TEXT_DOMAIN, "cannot get history for pool " "'%s'"), zhp->zpool_name)); case ENOTSUP: return (zfs_error_fmt(hdl, EZFS_BADVERSION, dgettext(TEXT_DOMAIN, "cannot get history for pool " "'%s', pool must be upgraded"), zhp->zpool_name)); default: return (zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get history for '%s'"), zhp->zpool_name)); } } *len = zc.zc_history_len; *off = zc.zc_history_offset; return (0); } /* * Retrieve the command history of a pool. */ int zpool_get_history(zpool_handle_t *zhp, nvlist_t **nvhisp, uint64_t *off, boolean_t *eof) { libzfs_handle_t *hdl = zhp->zpool_hdl; char *buf; int buflen = 128 * 1024; nvlist_t **records = NULL; uint_t numrecords = 0; int err = 0, i; uint64_t start = *off; buf = zfs_alloc(hdl, buflen); /* process about 1MiB a time */ while (*off - start < 1024 * 1024) { uint64_t bytes_read = buflen; uint64_t leftover; if ((err = get_history(zhp, buf, off, &bytes_read)) != 0) break; /* if nothing else was read in, we're at EOF, just return */ if (!bytes_read) { *eof = B_TRUE; break; } if ((err = zpool_history_unpack(buf, bytes_read, &leftover, &records, &numrecords)) != 0) { zpool_standard_error_fmt(hdl, err, dgettext(TEXT_DOMAIN, "cannot get history for '%s'"), zhp->zpool_name); break; } *off -= leftover; if (leftover == bytes_read) { /* * no progress made, because buffer is not big enough * to hold this record; resize and retry. */ buflen *= 2; free(buf); buf = zfs_alloc(hdl, buflen); } } free(buf); if (!err) { *nvhisp = fnvlist_alloc(); fnvlist_add_nvlist_array(*nvhisp, ZPOOL_HIST_RECORD, (const nvlist_t **)records, numrecords); } for (i = 0; i < numrecords; i++) nvlist_free(records[i]); free(records); return (err); } /* * Retrieve the next event given the passed 'zevent_fd' file descriptor. * If there is a new event available 'nvp' will contain a newly allocated * nvlist and 'dropped' will be set to the number of missed events since * the last call to this function. When 'nvp' is set to NULL it indicates * no new events are available. In either case the function returns 0 and * it is up to the caller to free 'nvp'. In the case of a fatal error the * function will return a non-zero value. When the function is called in * blocking mode (the default, unless the ZEVENT_NONBLOCK flag is passed), * it will not return until a new event is available. */ int zpool_events_next(libzfs_handle_t *hdl, nvlist_t **nvp, int *dropped, unsigned flags, int zevent_fd) { zfs_cmd_t zc = {"\0"}; int error = 0; *nvp = NULL; *dropped = 0; zc.zc_cleanup_fd = zevent_fd; if (flags & ZEVENT_NONBLOCK) zc.zc_guid = ZEVENT_NONBLOCK; zcmd_alloc_dst_nvlist(hdl, &zc, ZEVENT_SIZE); retry: if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_NEXT, &zc) != 0) { switch (errno) { case ESHUTDOWN: error = zfs_error_fmt(hdl, EZFS_POOLUNAVAIL, dgettext(TEXT_DOMAIN, "zfs shutdown")); goto out; case ENOENT: /* Blocking error case should not occur */ if (!(flags & ZEVENT_NONBLOCK)) error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); goto out; case ENOMEM: zcmd_expand_dst_nvlist(hdl, &zc); goto retry; default: error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); goto out; } } error = zcmd_read_dst_nvlist(hdl, &zc, nvp); if (error != 0) goto out; *dropped = (int)zc.zc_cookie; out: zcmd_free_nvlists(&zc); return (error); } /* * Clear all events. */ int zpool_events_clear(libzfs_handle_t *hdl, int *count) { zfs_cmd_t zc = {"\0"}; if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_CLEAR, &zc) != 0) return (zpool_standard_error(hdl, errno, dgettext(TEXT_DOMAIN, "cannot clear events"))); if (count != NULL) *count = (int)zc.zc_cookie; /* # of events cleared */ return (0); } /* * Seek to a specific EID, ZEVENT_SEEK_START, or ZEVENT_SEEK_END for * the passed zevent_fd file handle. On success zero is returned, * otherwise -1 is returned and hdl->libzfs_error is set to the errno. */ int zpool_events_seek(libzfs_handle_t *hdl, uint64_t eid, int zevent_fd) { zfs_cmd_t zc = {"\0"}; int error = 0; zc.zc_guid = eid; zc.zc_cleanup_fd = zevent_fd; if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_SEEK, &zc) != 0) { switch (errno) { case ENOENT: error = zfs_error_fmt(hdl, EZFS_NOENT, dgettext(TEXT_DOMAIN, "cannot get event")); break; case ENOMEM: error = zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot get event")); break; default: error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); break; } } return (error); } static void zpool_obj_to_path_impl(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj, char *pathname, size_t len, boolean_t always_unmounted) { zfs_cmd_t zc = {"\0"}; boolean_t mounted = B_FALSE; char *mntpnt = NULL; char dsname[ZFS_MAX_DATASET_NAME_LEN]; if (dsobj == 0) { /* special case for the MOS */ (void) snprintf(pathname, len, ":<0x%llx>", (longlong_t)obj); return; } /* get the dataset's name */ (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_obj = dsobj; if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_DSOBJ_TO_DSNAME, &zc) != 0) { /* just write out a path of two object numbers */ (void) snprintf(pathname, len, "<0x%llx>:<0x%llx>", (longlong_t)dsobj, (longlong_t)obj); return; } (void) strlcpy(dsname, zc.zc_value, sizeof (dsname)); /* find out if the dataset is mounted */ mounted = !always_unmounted && is_mounted(zhp->zpool_hdl, dsname, &mntpnt); /* get the corrupted object's path */ (void) strlcpy(zc.zc_name, dsname, sizeof (zc.zc_name)); zc.zc_obj = obj; if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_OBJ_TO_PATH, &zc) == 0) { if (mounted) { (void) snprintf(pathname, len, "%s%s", mntpnt, zc.zc_value); } else { (void) snprintf(pathname, len, "%s:%s", dsname, zc.zc_value); } } else { (void) snprintf(pathname, len, "%s:<0x%llx>", dsname, (longlong_t)obj); } free(mntpnt); } void zpool_obj_to_path(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj, char *pathname, size_t len) { zpool_obj_to_path_impl(zhp, dsobj, obj, pathname, len, B_FALSE); } void zpool_obj_to_path_ds(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj, char *pathname, size_t len) { zpool_obj_to_path_impl(zhp, dsobj, obj, pathname, len, B_TRUE); } /* * Wait while the specified activity is in progress in the pool. */ int zpool_wait(zpool_handle_t *zhp, zpool_wait_activity_t activity) { boolean_t missing; int error = zpool_wait_status(zhp, activity, &missing, NULL); if (missing) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, ENOENT, dgettext(TEXT_DOMAIN, "error waiting in pool '%s'"), zhp->zpool_name); return (ENOENT); } else { return (error); } } /* * Wait for the given activity and return the status of the wait (whether or not * any waiting was done) in the 'waited' parameter. Non-existent pools are * reported via the 'missing' parameter, rather than by printing an error * message. This is convenient when this function is called in a loop over a * long period of time (as it is, for example, by zpool's wait cmd). In that * scenario, a pool being exported or destroyed should be considered a normal * event, so we don't want to print an error when we find that the pool doesn't * exist. */ int zpool_wait_status(zpool_handle_t *zhp, zpool_wait_activity_t activity, boolean_t *missing, boolean_t *waited) { int error = lzc_wait(zhp->zpool_name, activity, waited); *missing = (error == ENOENT); if (*missing) return (0); if (error != 0) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, error, dgettext(TEXT_DOMAIN, "error waiting in pool '%s'"), zhp->zpool_name); } return (error); } int zpool_set_bootenv(zpool_handle_t *zhp, const nvlist_t *envmap) { int error = lzc_set_bootenv(zhp->zpool_name, envmap); if (error != 0) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, error, dgettext(TEXT_DOMAIN, "error setting bootenv in pool '%s'"), zhp->zpool_name); } return (error); } int zpool_get_bootenv(zpool_handle_t *zhp, nvlist_t **nvlp) { nvlist_t *nvl; int error; nvl = NULL; error = lzc_get_bootenv(zhp->zpool_name, &nvl); if (error != 0) { (void) zpool_standard_error_fmt(zhp->zpool_hdl, error, dgettext(TEXT_DOMAIN, "error getting bootenv in pool '%s'"), zhp->zpool_name); } else { *nvlp = nvl; } return (error); } /* * Attempt to read and parse feature file(s) (from "compatibility" property). * Files contain zpool feature names, comma or whitespace-separated. * Comments (# character to next newline) are discarded. * * Arguments: * compatibility : string containing feature filenames * features : either NULL or pointer to array of boolean * report : either NULL or pointer to string buffer * rlen : length of "report" buffer * * compatibility is NULL (unset), "", "off", "legacy", or list of * comma-separated filenames. filenames should either be absolute, * or relative to: * 1) ZPOOL_SYSCONF_COMPAT_D (eg: /etc/zfs/compatibility.d) or * 2) ZPOOL_DATA_COMPAT_D (eg: /usr/share/zfs/compatibility.d). * (Unset), "" or "off" => enable all features * "legacy" => disable all features * * Any feature names read from files which match unames in spa_feature_table * will have the corresponding boolean set in the features array (if non-NULL). * If more than one feature set specified, only features present in *all* of * them will be set. * * "report" if not NULL will be populated with a suitable status message. * * Return values: * ZPOOL_COMPATIBILITY_OK : files read and parsed ok * ZPOOL_COMPATIBILITY_BADFILE : file too big or not a text file * ZPOOL_COMPATIBILITY_BADTOKEN : SYSCONF file contains invalid feature name * ZPOOL_COMPATIBILITY_WARNTOKEN : DATA file contains invalid feature name * ZPOOL_COMPATIBILITY_NOFILES : no feature files found */ zpool_compat_status_t zpool_load_compat(const char *compat, boolean_t *features, char *report, size_t rlen) { int sdirfd, ddirfd, featfd; struct stat fs; char *fc; char *ps, *ls, *ws; char *file, *line, *word; char l_compat[ZFS_MAXPROPLEN]; boolean_t ret_nofiles = B_TRUE; boolean_t ret_badfile = B_FALSE; boolean_t ret_badtoken = B_FALSE; boolean_t ret_warntoken = B_FALSE; /* special cases (unset), "" and "off" => enable all features */ if (compat == NULL || compat[0] == '\0' || strcmp(compat, ZPOOL_COMPAT_OFF) == 0) { if (features != NULL) for (uint_t i = 0; i < SPA_FEATURES; i++) features[i] = B_TRUE; if (report != NULL) strlcpy(report, gettext("all features enabled"), rlen); return (ZPOOL_COMPATIBILITY_OK); } /* Final special case "legacy" => disable all features */ if (strcmp(compat, ZPOOL_COMPAT_LEGACY) == 0) { if (features != NULL) for (uint_t i = 0; i < SPA_FEATURES; i++) features[i] = B_FALSE; if (report != NULL) strlcpy(report, gettext("all features disabled"), rlen); return (ZPOOL_COMPATIBILITY_OK); } /* * Start with all true; will be ANDed with results from each file */ if (features != NULL) for (uint_t i = 0; i < SPA_FEATURES; i++) features[i] = B_TRUE; char err_badfile[ZFS_MAXPROPLEN] = ""; char err_badtoken[ZFS_MAXPROPLEN] = ""; /* * We ignore errors from the directory open() * as they're only needed if the filename is relative * which will be checked during the openat(). */ /* O_PATH safer than O_RDONLY if system allows it */ #if defined(O_PATH) #define ZC_DIR_FLAGS (O_DIRECTORY | O_CLOEXEC | O_PATH) #else #define ZC_DIR_FLAGS (O_DIRECTORY | O_CLOEXEC | O_RDONLY) #endif sdirfd = open(ZPOOL_SYSCONF_COMPAT_D, ZC_DIR_FLAGS); ddirfd = open(ZPOOL_DATA_COMPAT_D, ZC_DIR_FLAGS); (void) strlcpy(l_compat, compat, ZFS_MAXPROPLEN); for (file = strtok_r(l_compat, ",", &ps); file != NULL; file = strtok_r(NULL, ",", &ps)) { boolean_t l_features[SPA_FEATURES]; enum { Z_SYSCONF, Z_DATA } source; /* try sysconfdir first, then datadir */ source = Z_SYSCONF; if ((featfd = openat(sdirfd, file, O_RDONLY | O_CLOEXEC)) < 0) { featfd = openat(ddirfd, file, O_RDONLY | O_CLOEXEC); source = Z_DATA; } /* File readable and correct size? */ if (featfd < 0 || fstat(featfd, &fs) < 0 || fs.st_size < 1 || fs.st_size > ZPOOL_COMPAT_MAXSIZE) { (void) close(featfd); strlcat(err_badfile, file, ZFS_MAXPROPLEN); strlcat(err_badfile, " ", ZFS_MAXPROPLEN); ret_badfile = B_TRUE; continue; } /* Prefault the file if system allows */ #if defined(MAP_POPULATE) #define ZC_MMAP_FLAGS (MAP_PRIVATE | MAP_POPULATE) #elif defined(MAP_PREFAULT_READ) #define ZC_MMAP_FLAGS (MAP_PRIVATE | MAP_PREFAULT_READ) #else #define ZC_MMAP_FLAGS (MAP_PRIVATE) #endif /* private mmap() so we can strtok safely */ fc = (char *)mmap(NULL, fs.st_size, PROT_READ | PROT_WRITE, ZC_MMAP_FLAGS, featfd, 0); (void) close(featfd); /* map ok, and last character == newline? */ if (fc == MAP_FAILED || fc[fs.st_size - 1] != '\n') { (void) munmap((void *) fc, fs.st_size); strlcat(err_badfile, file, ZFS_MAXPROPLEN); strlcat(err_badfile, " ", ZFS_MAXPROPLEN); ret_badfile = B_TRUE; continue; } ret_nofiles = B_FALSE; for (uint_t i = 0; i < SPA_FEATURES; i++) l_features[i] = B_FALSE; /* replace final newline with NULL to ensure string ends */ fc[fs.st_size - 1] = '\0'; for (line = strtok_r(fc, "\n", &ls); line != NULL; line = strtok_r(NULL, "\n", &ls)) { /* discard comments */ char *r = strchr(line, '#'); if (r != NULL) *r = '\0'; for (word = strtok_r(line, ", \t", &ws); word != NULL; word = strtok_r(NULL, ", \t", &ws)) { /* Find matching feature name */ uint_t f; for (f = 0; f < SPA_FEATURES; f++) { zfeature_info_t *fi = &spa_feature_table[f]; if (strcmp(word, fi->fi_uname) == 0) { l_features[f] = B_TRUE; break; } } if (f < SPA_FEATURES) continue; /* found an unrecognized word */ /* lightly sanitize it */ if (strlen(word) > 32) word[32] = '\0'; for (char *c = word; *c != '\0'; c++) if (!isprint(*c)) *c = '?'; strlcat(err_badtoken, word, ZFS_MAXPROPLEN); strlcat(err_badtoken, " ", ZFS_MAXPROPLEN); if (source == Z_SYSCONF) ret_badtoken = B_TRUE; else ret_warntoken = B_TRUE; } } (void) munmap((void *) fc, fs.st_size); if (features != NULL) for (uint_t i = 0; i < SPA_FEATURES; i++) features[i] &= l_features[i]; } (void) close(sdirfd); (void) close(ddirfd); /* Return the most serious error */ if (ret_badfile) { if (report != NULL) snprintf(report, rlen, gettext("could not read/" "parse feature file(s): %s"), err_badfile); return (ZPOOL_COMPATIBILITY_BADFILE); } if (ret_nofiles) { if (report != NULL) strlcpy(report, gettext("no valid compatibility files specified"), rlen); return (ZPOOL_COMPATIBILITY_NOFILES); } if (ret_badtoken) { if (report != NULL) snprintf(report, rlen, gettext("invalid feature " "name(s) in local compatibility files: %s"), err_badtoken); return (ZPOOL_COMPATIBILITY_BADTOKEN); } if (ret_warntoken) { if (report != NULL) snprintf(report, rlen, gettext("unrecognized feature " "name(s) in distribution compatibility files: %s"), err_badtoken); return (ZPOOL_COMPATIBILITY_WARNTOKEN); } if (report != NULL) strlcpy(report, gettext("compatibility set ok"), rlen); return (ZPOOL_COMPATIBILITY_OK); } static int zpool_vdev_guid(zpool_handle_t *zhp, const char *vdevname, uint64_t *vdev_guid) { nvlist_t *tgt; boolean_t avail_spare, l2cache; verify(zhp != NULL); if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "pool is in an unavailable state")); return (zfs_error(zhp->zpool_hdl, EZFS_POOLUNAVAIL, errbuf)); } if ((tgt = zpool_find_vdev(zhp, vdevname, &avail_spare, &l2cache, NULL)) == NULL) { char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "can not find %s in %s"), vdevname, zhp->zpool_name); return (zfs_error(zhp->zpool_hdl, EZFS_NODEVICE, errbuf)); } *vdev_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); return (0); } /* * Get a vdev property value for 'prop' and return the value in * a pre-allocated buffer. */ int zpool_get_vdev_prop_value(nvlist_t *nvprop, vdev_prop_t prop, char *prop_name, char *buf, size_t len, zprop_source_t *srctype, boolean_t literal) { nvlist_t *nv; const char *strval; uint64_t intval; zprop_source_t src = ZPROP_SRC_NONE; if (prop == VDEV_PROP_USERPROP) { /* user property, prop_name must contain the property name */ assert(prop_name != NULL); if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) { src = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); strval = fnvlist_lookup_string(nv, ZPROP_VALUE); } else { /* user prop not found */ return (-1); } (void) strlcpy(buf, strval, len); if (srctype) *srctype = src; return (0); } if (prop_name == NULL) prop_name = (char *)vdev_prop_to_name(prop); switch (vdev_prop_get_type(prop)) { case PROP_TYPE_STRING: if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) { src = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); strval = fnvlist_lookup_string(nv, ZPROP_VALUE); } else { src = ZPROP_SRC_DEFAULT; if ((strval = vdev_prop_default_string(prop)) == NULL) strval = "-"; } (void) strlcpy(buf, strval, len); break; case PROP_TYPE_NUMBER: if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) { src = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); intval = fnvlist_lookup_uint64(nv, ZPROP_VALUE); } else { src = ZPROP_SRC_DEFAULT; intval = vdev_prop_default_numeric(prop); } switch (prop) { case VDEV_PROP_ASIZE: case VDEV_PROP_PSIZE: case VDEV_PROP_SIZE: case VDEV_PROP_BOOTSIZE: case VDEV_PROP_ALLOCATED: case VDEV_PROP_FREE: case VDEV_PROP_READ_ERRORS: case VDEV_PROP_WRITE_ERRORS: case VDEV_PROP_CHECKSUM_ERRORS: case VDEV_PROP_INITIALIZE_ERRORS: case VDEV_PROP_OPS_NULL: case VDEV_PROP_OPS_READ: case VDEV_PROP_OPS_WRITE: case VDEV_PROP_OPS_FREE: case VDEV_PROP_OPS_CLAIM: case VDEV_PROP_OPS_TRIM: case VDEV_PROP_BYTES_NULL: case VDEV_PROP_BYTES_READ: case VDEV_PROP_BYTES_WRITE: case VDEV_PROP_BYTES_FREE: case VDEV_PROP_BYTES_CLAIM: case VDEV_PROP_BYTES_TRIM: if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) zfs_nicenum(intval, buf, len); } break; case VDEV_PROP_EXPANDSZ: if (intval == 0) { (void) strlcpy(buf, "-", len); } else if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) zfs_nicenum(intval, buf, len); } break; case VDEV_PROP_CAPACITY: if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) snprintf(buf, len, "%llu%%", (u_longlong_t)intval); } break; case VDEV_PROP_FRAGMENTATION: if (intval == UINT64_MAX) { (void) strlcpy(buf, "-", len); } else { (void) snprintf(buf, len, "%llu%%", (u_longlong_t)intval); } break; case VDEV_PROP_STATE: if (literal) { (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } else { (void) strlcpy(buf, zpool_state_to_name(intval, VDEV_AUX_NONE), len); } break; default: (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } break; case PROP_TYPE_INDEX: if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) { src = fnvlist_lookup_uint64(nv, ZPROP_SOURCE); intval = fnvlist_lookup_uint64(nv, ZPROP_VALUE); } else { src = ZPROP_SRC_DEFAULT; intval = vdev_prop_default_numeric(prop); } if (vdev_prop_index_to_string(prop, intval, (const char **)&strval) != 0) return (-1); (void) strlcpy(buf, strval, len); break; default: abort(); } if (srctype) *srctype = src; return (0); } /* * Get a vdev property value for 'prop_name' and return the value in * a pre-allocated buffer. */ int zpool_get_vdev_prop(zpool_handle_t *zhp, const char *vdevname, vdev_prop_t prop, char *prop_name, char *buf, size_t len, zprop_source_t *srctype, boolean_t literal) { nvlist_t *reqnvl, *reqprops; nvlist_t *retprops = NULL; uint64_t vdev_guid = 0; int ret; if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0) return (ret); if (nvlist_alloc(&reqnvl, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); if (nvlist_alloc(&reqprops, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); fnvlist_add_uint64(reqnvl, ZPOOL_VDEV_PROPS_GET_VDEV, vdev_guid); if (prop != VDEV_PROP_USERPROP) { /* prop_name overrides prop value */ if (prop_name != NULL) prop = vdev_name_to_prop(prop_name); else prop_name = (char *)vdev_prop_to_name(prop); assert(prop < VDEV_NUM_PROPS); } assert(prop_name != NULL); if (nvlist_add_uint64(reqprops, prop_name, prop) != 0) { nvlist_free(reqnvl); nvlist_free(reqprops); return (no_memory(zhp->zpool_hdl)); } fnvlist_add_nvlist(reqnvl, ZPOOL_VDEV_PROPS_GET_PROPS, reqprops); ret = lzc_get_vdev_prop(zhp->zpool_name, reqnvl, &retprops); if (ret == 0) { ret = zpool_get_vdev_prop_value(retprops, prop, prop_name, buf, len, srctype, literal); } else { char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot get vdev property %s from" " %s in %s"), prop_name, vdevname, zhp->zpool_name); (void) zpool_standard_error(zhp->zpool_hdl, ret, errbuf); } nvlist_free(reqnvl); nvlist_free(reqprops); nvlist_free(retprops); return (ret); } /* * Get all vdev properties */ int zpool_get_all_vdev_props(zpool_handle_t *zhp, const char *vdevname, nvlist_t **outnvl) { nvlist_t *nvl = NULL; uint64_t vdev_guid = 0; int ret; if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0) return (ret); if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); fnvlist_add_uint64(nvl, ZPOOL_VDEV_PROPS_GET_VDEV, vdev_guid); ret = lzc_get_vdev_prop(zhp->zpool_name, nvl, outnvl); nvlist_free(nvl); if (ret) { char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot get vdev properties for" " %s in %s"), vdevname, zhp->zpool_name); (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf); } return (ret); } /* * Set vdev property */ int zpool_set_vdev_prop(zpool_handle_t *zhp, const char *vdevname, const char *propname, const char *propval) { int ret; nvlist_t *nvl = NULL; nvlist_t *outnvl = NULL; nvlist_t *props; nvlist_t *realprops; prop_flags_t flags = { 0 }; uint64_t version; uint64_t vdev_guid; if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0) return (ret); if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); fnvlist_add_uint64(nvl, ZPOOL_VDEV_PROPS_SET_VDEV, vdev_guid); if (nvlist_add_string(props, propname, propval) != 0) { nvlist_free(props); return (no_memory(zhp->zpool_hdl)); } char errbuf[ERRBUFLEN]; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot set property %s for %s on %s"), propname, vdevname, zhp->zpool_name); flags.vdevprop = 1; version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if ((realprops = zpool_valid_proplist(zhp->zpool_hdl, zhp->zpool_name, props, version, flags, errbuf)) == NULL) { nvlist_free(props); nvlist_free(nvl); return (-1); } nvlist_free(props); props = realprops; fnvlist_add_nvlist(nvl, ZPOOL_VDEV_PROPS_SET_PROPS, props); ret = lzc_set_vdev_prop(zhp->zpool_name, nvl, &outnvl); nvlist_free(props); nvlist_free(nvl); nvlist_free(outnvl); if (ret) (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf); return (ret); } diff --git a/module/os/linux/spl/spl-generic.c b/module/os/linux/spl/spl-generic.c index de91c44257aa..bc39ece9a427 100644 --- a/module/os/linux/spl/spl-generic.c +++ b/module/os/linux/spl/spl-generic.c @@ -1,834 +1,867 @@ /* * Copyright (C) 2007-2010 Lawrence Livermore National Security, LLC. * Copyright (C) 2007 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). * Written by Brian Behlendorf . * UCRL-CODE-235197 * * This file is part of the SPL, Solaris Porting Layer. * * The SPL is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation; either version 2 of the License, or (at your * option) any later version. * * The SPL is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with the SPL. If not, see . * * Solaris Porting Layer (SPL) Generic Implementation. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include +#include unsigned long spl_hostid = 0; EXPORT_SYMBOL(spl_hostid); /* CSTYLED */ module_param(spl_hostid, ulong, 0644); MODULE_PARM_DESC(spl_hostid, "The system hostid."); proc_t p0; EXPORT_SYMBOL(p0); /* * Xorshift Pseudo Random Number Generator based on work by Sebastiano Vigna * * "Further scramblings of Marsaglia's xorshift generators" * http://vigna.di.unimi.it/ftp/papers/xorshiftplus.pdf * * random_get_pseudo_bytes() is an API function on Illumos whose sole purpose * is to provide bytes containing random numbers. It is mapped to /dev/urandom * on Illumos, which uses a "FIPS 186-2 algorithm". No user of the SPL's * random_get_pseudo_bytes() needs bytes that are of cryptographic quality, so * we can implement it using a fast PRNG that we seed using Linux' actual * equivalent to random_get_pseudo_bytes(). We do this by providing each CPU * with an independent seed so that all calls to random_get_pseudo_bytes() are * free of atomic instructions. * * A consequence of using a fast PRNG is that using random_get_pseudo_bytes() * to generate words larger than 128 bits will paradoxically be limited to * `2^128 - 1` possibilities. This is because we have a sequence of `2^128 - 1` * 128-bit words and selecting the first will implicitly select the second. If * a caller finds this behavior undesirable, random_get_bytes() should be used * instead. * * XXX: Linux interrupt handlers that trigger within the critical section * formed by `s[1] = xp[1];` and `xp[0] = s[0];` and call this function will * see the same numbers. Nothing in the code currently calls this in an * interrupt handler, so this is considered to be okay. If that becomes a * problem, we could create a set of per-cpu variables for interrupt handlers * and use them when in_interrupt() from linux/preempt_mask.h evaluates to * true. */ void __percpu *spl_pseudo_entropy; /* * spl_rand_next()/spl_rand_jump() are copied from the following CC-0 licensed * file: * * http://xorshift.di.unimi.it/xorshift128plus.c */ static inline uint64_t spl_rand_next(uint64_t *s) { uint64_t s1 = s[0]; const uint64_t s0 = s[1]; s[0] = s0; s1 ^= s1 << 23; // a s[1] = s1 ^ s0 ^ (s1 >> 18) ^ (s0 >> 5); // b, c return (s[1] + s0); } static inline void spl_rand_jump(uint64_t *s) { static const uint64_t JUMP[] = { 0x8a5cd789635d2dff, 0x121fd2155c472f96 }; uint64_t s0 = 0; uint64_t s1 = 0; int i, b; for (i = 0; i < sizeof (JUMP) / sizeof (*JUMP); i++) for (b = 0; b < 64; b++) { if (JUMP[i] & 1ULL << b) { s0 ^= s[0]; s1 ^= s[1]; } (void) spl_rand_next(s); } s[0] = s0; s[1] = s1; } int random_get_pseudo_bytes(uint8_t *ptr, size_t len) { uint64_t *xp, s[2]; ASSERT(ptr); xp = get_cpu_ptr(spl_pseudo_entropy); s[0] = xp[0]; s[1] = xp[1]; while (len) { union { uint64_t ui64; uint8_t byte[sizeof (uint64_t)]; }entropy; int i = MIN(len, sizeof (uint64_t)); len -= i; entropy.ui64 = spl_rand_next(s); while (i--) *ptr++ = entropy.byte[i]; } xp[0] = s[0]; xp[1] = s[1]; put_cpu_ptr(spl_pseudo_entropy); return (0); } EXPORT_SYMBOL(random_get_pseudo_bytes); #if BITS_PER_LONG == 32 /* * Support 64/64 => 64 division on a 32-bit platform. While the kernel * provides a div64_u64() function for this we do not use it because the * implementation is flawed. There are cases which return incorrect * results as late as linux-2.6.35. Until this is fixed upstream the * spl must provide its own implementation. * * This implementation is a slightly modified version of the algorithm * proposed by the book 'Hacker's Delight'. The original source can be * found here and is available for use without restriction. * * http://www.hackersdelight.org/HDcode/newCode/divDouble.c */ /* * Calculate number of leading of zeros for a 64-bit value. */ static int nlz64(uint64_t x) { register int n = 0; if (x == 0) return (64); if (x <= 0x00000000FFFFFFFFULL) { n = n + 32; x = x << 32; } if (x <= 0x0000FFFFFFFFFFFFULL) { n = n + 16; x = x << 16; } if (x <= 0x00FFFFFFFFFFFFFFULL) { n = n + 8; x = x << 8; } if (x <= 0x0FFFFFFFFFFFFFFFULL) { n = n + 4; x = x << 4; } if (x <= 0x3FFFFFFFFFFFFFFFULL) { n = n + 2; x = x << 2; } if (x <= 0x7FFFFFFFFFFFFFFFULL) { n = n + 1; } return (n); } /* * Newer kernels have a div_u64() function but we define our own * to simplify portability between kernel versions. */ static inline uint64_t __div_u64(uint64_t u, uint32_t v) { (void) do_div(u, v); return (u); } /* * Turn off missing prototypes warning for these functions. They are * replacements for libgcc-provided functions and will never be called * directly. */ #pragma GCC diagnostic push #pragma GCC diagnostic ignored "-Wmissing-prototypes" /* * Implementation of 64-bit unsigned division for 32-bit machines. * * First the procedure takes care of the case in which the divisor is a * 32-bit quantity. There are two subcases: (1) If the left half of the * dividend is less than the divisor, one execution of do_div() is all that * is required (overflow is not possible). (2) Otherwise it does two * divisions, using the grade school method. */ uint64_t __udivdi3(uint64_t u, uint64_t v) { uint64_t u0, u1, v1, q0, q1, k; int n; if (v >> 32 == 0) { // If v < 2**32: if (u >> 32 < v) { // If u/v cannot overflow, return (__div_u64(u, v)); // just do one division. } else { // If u/v would overflow: u1 = u >> 32; // Break u into two halves. u0 = u & 0xFFFFFFFF; q1 = __div_u64(u1, v); // First quotient digit. k = u1 - q1 * v; // First remainder, < v. u0 += (k << 32); q0 = __div_u64(u0, v); // Seconds quotient digit. return ((q1 << 32) + q0); } } else { // If v >= 2**32: n = nlz64(v); // 0 <= n <= 31. v1 = (v << n) >> 32; // Normalize divisor, MSB is 1. u1 = u >> 1; // To ensure no overflow. q1 = __div_u64(u1, v1); // Get quotient from q0 = (q1 << n) >> 31; // Undo normalization and // division of u by 2. if (q0 != 0) // Make q0 correct or q0 = q0 - 1; // too small by 1. if ((u - q0 * v) >= v) q0 = q0 + 1; // Now q0 is correct. return (q0); } } EXPORT_SYMBOL(__udivdi3); #ifndef abs64 /* CSTYLED */ #define abs64(x) ({ uint64_t t = (x) >> 63; ((x) ^ t) - t; }) #endif /* * Implementation of 64-bit signed division for 32-bit machines. */ int64_t __divdi3(int64_t u, int64_t v) { int64_t q, t; q = __udivdi3(abs64(u), abs64(v)); t = (u ^ v) >> 63; // If u, v have different return ((q ^ t) - t); // signs, negate q. } EXPORT_SYMBOL(__divdi3); /* * Implementation of 64-bit unsigned modulo for 32-bit machines. */ uint64_t __umoddi3(uint64_t dividend, uint64_t divisor) { return (dividend - (divisor * __udivdi3(dividend, divisor))); } EXPORT_SYMBOL(__umoddi3); /* 64-bit signed modulo for 32-bit machines. */ int64_t __moddi3(int64_t n, int64_t d) { int64_t q; boolean_t nn = B_FALSE; if (n < 0) { nn = B_TRUE; n = -n; } if (d < 0) d = -d; q = __umoddi3(n, d); return (nn ? -q : q); } EXPORT_SYMBOL(__moddi3); /* * Implementation of 64-bit unsigned division/modulo for 32-bit machines. */ uint64_t __udivmoddi4(uint64_t n, uint64_t d, uint64_t *r) { uint64_t q = __udivdi3(n, d); if (r) *r = n - d * q; return (q); } EXPORT_SYMBOL(__udivmoddi4); /* * Implementation of 64-bit signed division/modulo for 32-bit machines. */ int64_t __divmoddi4(int64_t n, int64_t d, int64_t *r) { int64_t q, rr; boolean_t nn = B_FALSE; boolean_t nd = B_FALSE; if (n < 0) { nn = B_TRUE; n = -n; } if (d < 0) { nd = B_TRUE; d = -d; } q = __udivmoddi4(n, d, (uint64_t *)&rr); if (nn != nd) q = -q; if (nn) rr = -rr; if (r) *r = rr; return (q); } EXPORT_SYMBOL(__divmoddi4); #if defined(__arm) || defined(__arm__) /* * Implementation of 64-bit (un)signed division for 32-bit arm machines. * * Run-time ABI for the ARM Architecture (page 20). A pair of (unsigned) * long longs is returned in {{r0, r1}, {r2,r3}}, the quotient in {r0, r1}, * and the remainder in {r2, r3}. The return type is specifically left * set to 'void' to ensure the compiler does not overwrite these registers * during the return. All results are in registers as per ABI */ void __aeabi_uldivmod(uint64_t u, uint64_t v) { uint64_t res; uint64_t mod; res = __udivdi3(u, v); mod = __umoddi3(u, v); { register uint32_t r0 asm("r0") = (res & 0xFFFFFFFF); register uint32_t r1 asm("r1") = (res >> 32); register uint32_t r2 asm("r2") = (mod & 0xFFFFFFFF); register uint32_t r3 asm("r3") = (mod >> 32); asm volatile("" : "+r"(r0), "+r"(r1), "+r"(r2), "+r"(r3) /* output */ : "r"(r0), "r"(r1), "r"(r2), "r"(r3)); /* input */ return; /* r0; */ } } EXPORT_SYMBOL(__aeabi_uldivmod); void __aeabi_ldivmod(int64_t u, int64_t v) { int64_t res; uint64_t mod; res = __divdi3(u, v); mod = __umoddi3(u, v); { register uint32_t r0 asm("r0") = (res & 0xFFFFFFFF); register uint32_t r1 asm("r1") = (res >> 32); register uint32_t r2 asm("r2") = (mod & 0xFFFFFFFF); register uint32_t r3 asm("r3") = (mod >> 32); asm volatile("" : "+r"(r0), "+r"(r1), "+r"(r2), "+r"(r3) /* output */ : "r"(r0), "r"(r1), "r"(r2), "r"(r3)); /* input */ return; /* r0; */ } } EXPORT_SYMBOL(__aeabi_ldivmod); #endif /* __arm || __arm__ */ #pragma GCC diagnostic pop #endif /* BITS_PER_LONG */ /* * NOTE: The strtoxx behavior is solely based on my reading of the Solaris * ddi_strtol(9F) man page. I have not verified the behavior of these * functions against their Solaris counterparts. It is possible that I * may have misinterpreted the man page or the man page is incorrect. */ int ddi_strtol(const char *, char **, int, long *); int ddi_strtoull(const char *, char **, int, unsigned long long *); int ddi_strtoll(const char *, char **, int, long long *); #define define_ddi_strtox(type, valtype) \ int ddi_strto##type(const char *str, char **endptr, \ int base, valtype *result) \ { \ valtype last_value, value = 0; \ char *ptr = (char *)str; \ int digit, minus = 0; \ \ while (strchr(" \t\n\r\f", *ptr)) \ ++ptr; \ \ if (strlen(ptr) == 0) \ return (EINVAL); \ \ switch (*ptr) { \ case '-': \ minus = 1; \ zfs_fallthrough; \ case '+': \ ++ptr; \ break; \ } \ \ /* Auto-detect base based on prefix */ \ if (!base) { \ if (str[0] == '0') { \ if (tolower(str[1]) == 'x' && isxdigit(str[2])) { \ base = 16; /* hex */ \ ptr += 2; \ } else if (str[1] >= '0' && str[1] < 8) { \ base = 8; /* octal */ \ ptr += 1; \ } else { \ return (EINVAL); \ } \ } else { \ base = 10; /* decimal */ \ } \ } \ \ while (1) { \ if (isdigit(*ptr)) \ digit = *ptr - '0'; \ else if (isalpha(*ptr)) \ digit = tolower(*ptr) - 'a' + 10; \ else \ break; \ \ if (digit >= base) \ break; \ \ last_value = value; \ value = value * base + digit; \ if (last_value > value) /* Overflow */ \ return (ERANGE); \ \ ptr++; \ } \ \ *result = minus ? -value : value; \ \ if (endptr) \ *endptr = ptr; \ \ return (0); \ } \ define_ddi_strtox(l, long) define_ddi_strtox(ull, unsigned long long) define_ddi_strtox(ll, long long) EXPORT_SYMBOL(ddi_strtol); EXPORT_SYMBOL(ddi_strtoll); EXPORT_SYMBOL(ddi_strtoull); int ddi_copyin(const void *from, void *to, size_t len, int flags) { /* Fake ioctl() issued by kernel, 'from' is a kernel address */ if (flags & FKIOCTL) { memcpy(to, from, len); return (0); } return (copyin(from, to, len)); } EXPORT_SYMBOL(ddi_copyin); +/* + * Post a uevent to userspace whenever a new vdev adds to the pool. It is + * necessary to sync blkid information with udev, which zed daemon uses + * during device hotplug to identify the vdev. + */ +void +spl_signal_kobj_evt(struct block_device *bdev) +{ +#if defined(HAVE_BDEV_KOBJ) || defined(HAVE_PART_TO_DEV) +#ifdef HAVE_BDEV_KOBJ + struct kobject *disk_kobj = bdev_kobj(bdev); +#else + struct kobject *disk_kobj = &part_to_dev(bdev->bd_part)->kobj; +#endif + if (disk_kobj) { + int ret = kobject_uevent(disk_kobj, KOBJ_CHANGE); + if (ret) { + pr_warn("ZFS: Sending event '%d' to kobject: '%s'" + " (%p): failed(ret:%d)\n", KOBJ_CHANGE, + kobject_name(disk_kobj), disk_kobj, ret); + } + } +#else +/* + * This is encountered if neither bdev_kobj() nor part_to_dev() is available + * in the kernel - likely due to an API change that needs to be chased down. + */ +#error "Unsupported kernel: unable to get struct kobj from bdev" +#endif +} +EXPORT_SYMBOL(spl_signal_kobj_evt); + int ddi_copyout(const void *from, void *to, size_t len, int flags) { /* Fake ioctl() issued by kernel, 'from' is a kernel address */ if (flags & FKIOCTL) { memcpy(to, from, len); return (0); } return (copyout(from, to, len)); } EXPORT_SYMBOL(ddi_copyout); static ssize_t spl_kernel_read(struct file *file, void *buf, size_t count, loff_t *pos) { #if defined(HAVE_KERNEL_READ_PPOS) return (kernel_read(file, buf, count, pos)); #else mm_segment_t saved_fs; ssize_t ret; saved_fs = get_fs(); set_fs(KERNEL_DS); ret = vfs_read(file, (void __user *)buf, count, pos); set_fs(saved_fs); return (ret); #endif } static int spl_getattr(struct file *filp, struct kstat *stat) { int rc; ASSERT(filp); ASSERT(stat); #if defined(HAVE_4ARGS_VFS_GETATTR) rc = vfs_getattr(&filp->f_path, stat, STATX_BASIC_STATS, AT_STATX_SYNC_AS_STAT); #elif defined(HAVE_2ARGS_VFS_GETATTR) rc = vfs_getattr(&filp->f_path, stat); #elif defined(HAVE_3ARGS_VFS_GETATTR) rc = vfs_getattr(filp->f_path.mnt, filp->f_dentry, stat); #else #error "No available vfs_getattr()" #endif if (rc) return (-rc); return (0); } /* * Read the unique system identifier from the /etc/hostid file. * * The behavior of /usr/bin/hostid on Linux systems with the * regular eglibc and coreutils is: * * 1. Generate the value if the /etc/hostid file does not exist * or if the /etc/hostid file is less than four bytes in size. * * 2. If the /etc/hostid file is at least 4 bytes, then return * the first four bytes [0..3] in native endian order. * * 3. Always ignore bytes [4..] if they exist in the file. * * Only the first four bytes are significant, even on systems that * have a 64-bit word size. * * See: * * eglibc: sysdeps/unix/sysv/linux/gethostid.c * coreutils: src/hostid.c * * Notes: * * The /etc/hostid file on Solaris is a text file that often reads: * * # DO NOT EDIT * "0123456789" * * Directly copying this file to Linux results in a constant * hostid of 4f442023 because the default comment constitutes * the first four bytes of the file. * */ static char *spl_hostid_path = HW_HOSTID_PATH; module_param(spl_hostid_path, charp, 0444); MODULE_PARM_DESC(spl_hostid_path, "The system hostid file (/etc/hostid)"); static int hostid_read(uint32_t *hostid) { uint64_t size; uint32_t value = 0; int error; loff_t off; struct file *filp; struct kstat stat; filp = filp_open(spl_hostid_path, 0, 0); if (IS_ERR(filp)) return (ENOENT); error = spl_getattr(filp, &stat); if (error) { filp_close(filp, 0); return (error); } size = stat.size; // cppcheck-suppress sizeofwithnumericparameter if (size < sizeof (HW_HOSTID_MASK)) { filp_close(filp, 0); return (EINVAL); } off = 0; /* * Read directly into the variable like eglibc does. * Short reads are okay; native behavior is preserved. */ error = spl_kernel_read(filp, &value, sizeof (value), &off); if (error < 0) { filp_close(filp, 0); return (EIO); } /* Mask down to 32 bits like coreutils does. */ *hostid = (value & HW_HOSTID_MASK); filp_close(filp, 0); return (0); } /* * Return the system hostid. Preferentially use the spl_hostid module option * when set, otherwise use the value in the /etc/hostid file. */ uint32_t zone_get_hostid(void *zone) { uint32_t hostid; ASSERT3P(zone, ==, NULL); if (spl_hostid != 0) return ((uint32_t)(spl_hostid & HW_HOSTID_MASK)); if (hostid_read(&hostid) == 0) return (hostid); return (0); } EXPORT_SYMBOL(zone_get_hostid); static int spl_kvmem_init(void) { int rc = 0; rc = spl_kmem_init(); if (rc) return (rc); rc = spl_vmem_init(); if (rc) { spl_kmem_fini(); return (rc); } return (rc); } /* * We initialize the random number generator with 128 bits of entropy from the * system random number generator. In the improbable case that we have a zero * seed, we fallback to the system jiffies, unless it is also zero, in which * situation we use a preprogrammed seed. We step forward by 2^64 iterations to * initialize each of the per-cpu seeds so that the sequences generated on each * CPU are guaranteed to never overlap in practice. */ static int __init spl_random_init(void) { uint64_t s[2]; int i = 0; spl_pseudo_entropy = __alloc_percpu(2 * sizeof (uint64_t), sizeof (uint64_t)); if (!spl_pseudo_entropy) return (-ENOMEM); get_random_bytes(s, sizeof (s)); if (s[0] == 0 && s[1] == 0) { if (jiffies != 0) { s[0] = jiffies; s[1] = ~0 - jiffies; } else { (void) memcpy(s, "improbable seed", sizeof (s)); } printk("SPL: get_random_bytes() returned 0 " "when generating random seed. Setting initial seed to " "0x%016llx%016llx.\n", cpu_to_be64(s[0]), cpu_to_be64(s[1])); } for_each_possible_cpu(i) { uint64_t *wordp = per_cpu_ptr(spl_pseudo_entropy, i); spl_rand_jump(s); wordp[0] = s[0]; wordp[1] = s[1]; } return (0); } static void spl_random_fini(void) { free_percpu(spl_pseudo_entropy); } static void spl_kvmem_fini(void) { spl_vmem_fini(); spl_kmem_fini(); } static int __init spl_init(void) { int rc = 0; if ((rc = spl_random_init())) goto out0; if ((rc = spl_kvmem_init())) goto out1; if ((rc = spl_tsd_init())) goto out2; if ((rc = spl_taskq_init())) goto out3; if ((rc = spl_kmem_cache_init())) goto out4; if ((rc = spl_proc_init())) goto out5; if ((rc = spl_kstat_init())) goto out6; if ((rc = spl_zlib_init())) goto out7; if ((rc = spl_zone_init())) goto out8; return (rc); out8: spl_zlib_fini(); out7: spl_kstat_fini(); out6: spl_proc_fini(); out5: spl_kmem_cache_fini(); out4: spl_taskq_fini(); out3: spl_tsd_fini(); out2: spl_kvmem_fini(); out1: spl_random_fini(); out0: return (rc); } static void __exit spl_fini(void) { spl_zone_fini(); spl_zlib_fini(); spl_kstat_fini(); spl_proc_fini(); spl_kmem_cache_fini(); spl_taskq_fini(); spl_tsd_fini(); spl_kvmem_fini(); spl_random_fini(); } module_init(spl_init); module_exit(spl_fini); MODULE_DESCRIPTION("Solaris Porting Layer"); MODULE_AUTHOR(ZFS_META_AUTHOR); MODULE_LICENSE("GPL"); MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE); diff --git a/module/os/linux/zfs/vdev_disk.c b/module/os/linux/zfs/vdev_disk.c index ba7adcc1b576..0fed09df5203 100644 --- a/module/os/linux/zfs/vdev_disk.c +++ b/module/os/linux/zfs/vdev_disk.c @@ -1,1024 +1,1044 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (C) 2008-2010 Lawrence Livermore National Security, LLC. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). * Rewritten for Linux by Brian Behlendorf . * LLNL-CODE-403049. * Copyright (c) 2012, 2019 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #ifdef HAVE_LINUX_BLK_CGROUP_HEADER #include #endif typedef struct vdev_disk { struct block_device *vd_bdev; krwlock_t vd_lock; } vdev_disk_t; /* * Unique identifier for the exclusive vdev holder. */ static void *zfs_vdev_holder = VDEV_HOLDER; /* * Wait up to zfs_vdev_open_timeout_ms milliseconds before determining the * device is missing. The missing path may be transient since the links * can be briefly removed and recreated in response to udev events. */ static unsigned zfs_vdev_open_timeout_ms = 1000; /* * Size of the "reserved" partition, in blocks. */ #define EFI_MIN_RESV_SIZE (16 * 1024) /* * Virtual device vector for disks. */ typedef struct dio_request { zio_t *dr_zio; /* Parent ZIO */ atomic_t dr_ref; /* References */ int dr_error; /* Bio error */ int dr_bio_count; /* Count of bio's */ struct bio *dr_bio[0]; /* Attached bio's */ } dio_request_t; static fmode_t vdev_bdev_mode(spa_mode_t spa_mode) { fmode_t mode = 0; if (spa_mode & SPA_MODE_READ) mode |= FMODE_READ; if (spa_mode & SPA_MODE_WRITE) mode |= FMODE_WRITE; return (mode); } /* * Returns the usable capacity (in bytes) for the partition or disk. */ static uint64_t bdev_capacity(struct block_device *bdev) { return (i_size_read(bdev->bd_inode)); } #if !defined(HAVE_BDEV_WHOLE) static inline struct block_device * bdev_whole(struct block_device *bdev) { return (bdev->bd_contains); } #endif #if defined(HAVE_BDEVNAME) #define vdev_bdevname(bdev, name) bdevname(bdev, name) #else static inline void vdev_bdevname(struct block_device *bdev, char *name) { snprintf(name, BDEVNAME_SIZE, "%pg", bdev); } #endif /* * Returns the maximum expansion capacity of the block device (in bytes). * * It is possible to expand a vdev when it has been created as a wholedisk * and the containing block device has increased in capacity. Or when the * partition containing the pool has been manually increased in size. * * This function is only responsible for calculating the potential expansion * size so it can be reported by 'zpool list'. The efi_use_whole_disk() is * responsible for verifying the expected partition layout in the wholedisk * case, and updating the partition table if appropriate. Once the partition * size has been increased the additional capacity will be visible using * bdev_capacity(). * * The returned maximum expansion capacity is always expected to be larger, or * at the very least equal, to its usable capacity to prevent overestimating * the pool expandsize. */ static uint64_t bdev_max_capacity(struct block_device *bdev, uint64_t wholedisk) { uint64_t psize; int64_t available; if (wholedisk && bdev != bdev_whole(bdev)) { /* * When reporting maximum expansion capacity for a wholedisk * deduct any capacity which is expected to be lost due to * alignment restrictions. Over reporting this value isn't * harmful and would only result in slightly less capacity * than expected post expansion. * The estimated available space may be slightly smaller than * bdev_capacity() for devices where the number of sectors is * not a multiple of the alignment size and the partition layout * is keeping less than PARTITION_END_ALIGNMENT bytes after the * "reserved" EFI partition: in such cases return the device * usable capacity. */ available = i_size_read(bdev_whole(bdev)->bd_inode) - ((EFI_MIN_RESV_SIZE + NEW_START_BLOCK + PARTITION_END_ALIGNMENT) << SECTOR_BITS); psize = MAX(available, bdev_capacity(bdev)); } else { psize = bdev_capacity(bdev); } return (psize); } static void vdev_disk_error(zio_t *zio) { /* * This function can be called in interrupt context, for instance while * handling IRQs coming from a misbehaving disk device; use printk() * which is safe from any context. */ printk(KERN_WARNING "zio pool=%s vdev=%s error=%d type=%d " "offset=%llu size=%llu flags=%x\n", spa_name(zio->io_spa), zio->io_vd->vdev_path, zio->io_error, zio->io_type, (u_longlong_t)zio->io_offset, (u_longlong_t)zio->io_size, zio->io_flags); } +static void +vdev_disk_kobj_evt_post(vdev_t *v) +{ + vdev_disk_t *vd = v->vdev_tsd; + if (vd && vd->vd_bdev) { + spl_signal_kobj_evt(vd->vd_bdev); + } else { + vdev_dbgmsg(v, "vdev_disk_t is NULL for VDEV:%s\n", + v->vdev_path); + } +} + static int vdev_disk_open(vdev_t *v, uint64_t *psize, uint64_t *max_psize, uint64_t *logical_ashift, uint64_t *physical_ashift) { struct block_device *bdev; fmode_t mode = vdev_bdev_mode(spa_mode(v->vdev_spa)); hrtime_t timeout = MSEC2NSEC(zfs_vdev_open_timeout_ms); vdev_disk_t *vd; /* Must have a pathname and it must be absolute. */ if (v->vdev_path == NULL || v->vdev_path[0] != '/') { v->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL; vdev_dbgmsg(v, "invalid vdev_path"); return (SET_ERROR(EINVAL)); } /* * Reopen the device if it is currently open. When expanding a * partition force re-scanning the partition table if userland * did not take care of this already. We need to do this while closed * in order to get an accurate updated block device size. Then * since udev may need to recreate the device links increase the * open retry timeout before reporting the device as unavailable. */ vd = v->vdev_tsd; if (vd) { char disk_name[BDEVNAME_SIZE + 6] = "/dev/"; boolean_t reread_part = B_FALSE; rw_enter(&vd->vd_lock, RW_WRITER); bdev = vd->vd_bdev; vd->vd_bdev = NULL; if (bdev) { if (v->vdev_expanding && bdev != bdev_whole(bdev)) { vdev_bdevname(bdev_whole(bdev), disk_name + 5); /* * If userland has BLKPG_RESIZE_PARTITION, * then it should have updated the partition * table already. We can detect this by * comparing our current physical size * with that of the device. If they are * the same, then we must not have * BLKPG_RESIZE_PARTITION or it failed to * update the partition table online. We * fallback to rescanning the partition * table from the kernel below. However, * if the capacity already reflects the * updated partition, then we skip * rescanning the partition table here. */ if (v->vdev_psize == bdev_capacity(bdev)) reread_part = B_TRUE; } blkdev_put(bdev, mode | FMODE_EXCL); } if (reread_part) { bdev = blkdev_get_by_path(disk_name, mode | FMODE_EXCL, zfs_vdev_holder); if (!IS_ERR(bdev)) { int error = vdev_bdev_reread_part(bdev); blkdev_put(bdev, mode | FMODE_EXCL); if (error == 0) { timeout = MSEC2NSEC( zfs_vdev_open_timeout_ms * 2); } } } } else { vd = kmem_zalloc(sizeof (vdev_disk_t), KM_SLEEP); rw_init(&vd->vd_lock, NULL, RW_DEFAULT, NULL); rw_enter(&vd->vd_lock, RW_WRITER); } /* * Devices are always opened by the path provided at configuration * time. This means that if the provided path is a udev by-id path * then drives may be re-cabled without an issue. If the provided * path is a udev by-path path, then the physical location information * will be preserved. This can be critical for more complicated * configurations where drives are located in specific physical * locations to maximize the systems tolerance to component failure. * * Alternatively, you can provide your own udev rule to flexibly map * the drives as you see fit. It is not advised that you use the * /dev/[hd]d devices which may be reordered due to probing order. * Devices in the wrong locations will be detected by the higher * level vdev validation. * * The specified paths may be briefly removed and recreated in * response to udev events. This should be exceptionally unlikely * because the zpool command makes every effort to verify these paths * have already settled prior to reaching this point. Therefore, * a ENOENT failure at this point is highly likely to be transient * and it is reasonable to sleep and retry before giving up. In * practice delays have been observed to be on the order of 100ms. * * When ERESTARTSYS is returned it indicates the block device is * a zvol which could not be opened due to the deadlock detection * logic in zvol_open(). Extend the timeout and retry the open * subsequent attempts are expected to eventually succeed. */ hrtime_t start = gethrtime(); bdev = ERR_PTR(-ENXIO); while (IS_ERR(bdev) && ((gethrtime() - start) < timeout)) { bdev = blkdev_get_by_path(v->vdev_path, mode | FMODE_EXCL, zfs_vdev_holder); if (unlikely(PTR_ERR(bdev) == -ENOENT)) { + /* + * There is no point of waiting since device is removed + * explicitly + */ + if (v->vdev_removed) + break; + schedule_timeout(MSEC_TO_TICK(10)); } else if (unlikely(PTR_ERR(bdev) == -ERESTARTSYS)) { timeout = MSEC2NSEC(zfs_vdev_open_timeout_ms * 10); continue; } else if (IS_ERR(bdev)) { break; } } if (IS_ERR(bdev)) { int error = -PTR_ERR(bdev); vdev_dbgmsg(v, "open error=%d timeout=%llu/%llu", error, (u_longlong_t)(gethrtime() - start), (u_longlong_t)timeout); vd->vd_bdev = NULL; v->vdev_tsd = vd; rw_exit(&vd->vd_lock); return (SET_ERROR(error)); } else { vd->vd_bdev = bdev; v->vdev_tsd = vd; rw_exit(&vd->vd_lock); } /* Determine the physical block size */ int physical_block_size = bdev_physical_block_size(vd->vd_bdev); /* Determine the logical block size */ int logical_block_size = bdev_logical_block_size(vd->vd_bdev); /* Clear the nowritecache bit, causes vdev_reopen() to try again. */ v->vdev_nowritecache = B_FALSE; /* Set when device reports it supports TRIM. */ v->vdev_has_trim = bdev_discard_supported(vd->vd_bdev); /* Set when device reports it supports secure TRIM. */ v->vdev_has_securetrim = bdev_secure_discard_supported(vd->vd_bdev); /* Inform the ZIO pipeline that we are non-rotational */ v->vdev_nonrot = blk_queue_nonrot(bdev_get_queue(vd->vd_bdev)); /* Physical volume size in bytes for the partition */ *psize = bdev_capacity(vd->vd_bdev); /* Physical volume size in bytes including possible expansion space */ *max_psize = bdev_max_capacity(vd->vd_bdev, v->vdev_wholedisk); /* Based on the minimum sector size set the block size */ *physical_ashift = highbit64(MAX(physical_block_size, SPA_MINBLOCKSIZE)) - 1; *logical_ashift = highbit64(MAX(logical_block_size, SPA_MINBLOCKSIZE)) - 1; return (0); } static void vdev_disk_close(vdev_t *v) { vdev_disk_t *vd = v->vdev_tsd; if (v->vdev_reopening || vd == NULL) return; if (vd->vd_bdev != NULL) { blkdev_put(vd->vd_bdev, vdev_bdev_mode(spa_mode(v->vdev_spa)) | FMODE_EXCL); } rw_destroy(&vd->vd_lock); kmem_free(vd, sizeof (vdev_disk_t)); v->vdev_tsd = NULL; } static dio_request_t * vdev_disk_dio_alloc(int bio_count) { dio_request_t *dr = kmem_zalloc(sizeof (dio_request_t) + sizeof (struct bio *) * bio_count, KM_SLEEP); atomic_set(&dr->dr_ref, 0); dr->dr_bio_count = bio_count; dr->dr_error = 0; for (int i = 0; i < dr->dr_bio_count; i++) dr->dr_bio[i] = NULL; return (dr); } static void vdev_disk_dio_free(dio_request_t *dr) { int i; for (i = 0; i < dr->dr_bio_count; i++) if (dr->dr_bio[i]) bio_put(dr->dr_bio[i]); kmem_free(dr, sizeof (dio_request_t) + sizeof (struct bio *) * dr->dr_bio_count); } static void vdev_disk_dio_get(dio_request_t *dr) { atomic_inc(&dr->dr_ref); } static int vdev_disk_dio_put(dio_request_t *dr) { int rc = atomic_dec_return(&dr->dr_ref); /* * Free the dio_request when the last reference is dropped and * ensure zio_interpret is called only once with the correct zio */ if (rc == 0) { zio_t *zio = dr->dr_zio; int error = dr->dr_error; vdev_disk_dio_free(dr); if (zio) { zio->io_error = error; ASSERT3S(zio->io_error, >=, 0); if (zio->io_error) vdev_disk_error(zio); zio_delay_interrupt(zio); } } return (rc); } BIO_END_IO_PROTO(vdev_disk_physio_completion, bio, error) { dio_request_t *dr = bio->bi_private; int rc; if (dr->dr_error == 0) { #ifdef HAVE_1ARG_BIO_END_IO_T dr->dr_error = BIO_END_IO_ERROR(bio); #else if (error) dr->dr_error = -(error); else if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) dr->dr_error = EIO; #endif } /* Drop reference acquired by __vdev_disk_physio */ rc = vdev_disk_dio_put(dr); } static inline void vdev_submit_bio_impl(struct bio *bio) { #ifdef HAVE_1ARG_SUBMIT_BIO (void) submit_bio(bio); #else (void) submit_bio(bio_data_dir(bio), bio); #endif } /* * preempt_schedule_notrace is GPL-only which breaks the ZFS build, so * replace it with preempt_schedule under the following condition: */ #if defined(CONFIG_ARM64) && \ defined(CONFIG_PREEMPTION) && \ defined(CONFIG_BLK_CGROUP) #define preempt_schedule_notrace(x) preempt_schedule(x) #endif /* * As for the Linux 5.18 kernel bio_alloc() expects a block_device struct * as an argument removing the need to set it with bio_set_dev(). This * removes the need for all of the following compatibility code. */ #if !defined(HAVE_BIO_ALLOC_4ARG) #ifdef HAVE_BIO_SET_DEV #if defined(CONFIG_BLK_CGROUP) && defined(HAVE_BIO_SET_DEV_GPL_ONLY) /* * The Linux 5.5 kernel updated percpu_ref_tryget() which is inlined by * blkg_tryget() to use rcu_read_lock() instead of rcu_read_lock_sched(). * As a side effect the function was converted to GPL-only. Define our * own version when needed which uses rcu_read_lock_sched(). * * The Linux 5.17 kernel split linux/blk-cgroup.h into a private and a public * part, moving blkg_tryget into the private one. Define our own version. */ #if defined(HAVE_BLKG_TRYGET_GPL_ONLY) || !defined(HAVE_BLKG_TRYGET) static inline bool vdev_blkg_tryget(struct blkcg_gq *blkg) { struct percpu_ref *ref = &blkg->refcnt; unsigned long __percpu *count; bool rc; rcu_read_lock_sched(); if (__ref_is_percpu(ref, &count)) { this_cpu_inc(*count); rc = true; } else { #ifdef ZFS_PERCPU_REF_COUNT_IN_DATA rc = atomic_long_inc_not_zero(&ref->data->count); #else rc = atomic_long_inc_not_zero(&ref->count); #endif } rcu_read_unlock_sched(); return (rc); } #else #define vdev_blkg_tryget(bg) blkg_tryget(bg) #endif #ifdef HAVE_BIO_SET_DEV_MACRO /* * The Linux 5.0 kernel updated the bio_set_dev() macro so it calls the * GPL-only bio_associate_blkg() symbol thus inadvertently converting * the entire macro. Provide a minimal version which always assigns the * request queue's root_blkg to the bio. */ static inline void vdev_bio_associate_blkg(struct bio *bio) { #if defined(HAVE_BIO_BDEV_DISK) struct request_queue *q = bio->bi_bdev->bd_disk->queue; #else struct request_queue *q = bio->bi_disk->queue; #endif ASSERT3P(q, !=, NULL); ASSERT3P(bio->bi_blkg, ==, NULL); if (q->root_blkg && vdev_blkg_tryget(q->root_blkg)) bio->bi_blkg = q->root_blkg; } #define bio_associate_blkg vdev_bio_associate_blkg #else static inline void vdev_bio_set_dev(struct bio *bio, struct block_device *bdev) { #if defined(HAVE_BIO_BDEV_DISK) struct request_queue *q = bdev->bd_disk->queue; #else struct request_queue *q = bio->bi_disk->queue; #endif bio_clear_flag(bio, BIO_REMAPPED); if (bio->bi_bdev != bdev) bio_clear_flag(bio, BIO_THROTTLED); bio->bi_bdev = bdev; ASSERT3P(q, !=, NULL); ASSERT3P(bio->bi_blkg, ==, NULL); if (q->root_blkg && vdev_blkg_tryget(q->root_blkg)) bio->bi_blkg = q->root_blkg; } #define bio_set_dev vdev_bio_set_dev #endif #endif #else /* * Provide a bio_set_dev() helper macro for pre-Linux 4.14 kernels. */ static inline void bio_set_dev(struct bio *bio, struct block_device *bdev) { bio->bi_bdev = bdev; } #endif /* HAVE_BIO_SET_DEV */ #endif /* !HAVE_BIO_ALLOC_4ARG */ static inline void vdev_submit_bio(struct bio *bio) { struct bio_list *bio_list = current->bio_list; current->bio_list = NULL; vdev_submit_bio_impl(bio); current->bio_list = bio_list; } static inline struct bio * vdev_bio_alloc(struct block_device *bdev, gfp_t gfp_mask, unsigned short nr_vecs) { struct bio *bio; #ifdef HAVE_BIO_ALLOC_4ARG bio = bio_alloc(bdev, nr_vecs, 0, gfp_mask); #else bio = bio_alloc(gfp_mask, nr_vecs); if (likely(bio != NULL)) bio_set_dev(bio, bdev); #endif return (bio); } static inline unsigned int vdev_bio_max_segs(zio_t *zio, int bio_size, uint64_t abd_offset) { unsigned long nr_segs = abd_nr_pages_off(zio->io_abd, bio_size, abd_offset); #ifdef HAVE_BIO_MAX_SEGS return (bio_max_segs(nr_segs)); #else return (MIN(nr_segs, BIO_MAX_PAGES)); #endif } static int __vdev_disk_physio(struct block_device *bdev, zio_t *zio, size_t io_size, uint64_t io_offset, int rw, int flags) { dio_request_t *dr; uint64_t abd_offset; uint64_t bio_offset; int bio_size; int bio_count = 16; int error = 0; struct blk_plug plug; unsigned short nr_vecs; /* * Accessing outside the block device is never allowed. */ if (io_offset + io_size > bdev->bd_inode->i_size) { vdev_dbgmsg(zio->io_vd, "Illegal access %llu size %llu, device size %llu", (u_longlong_t)io_offset, (u_longlong_t)io_size, (u_longlong_t)i_size_read(bdev->bd_inode)); return (SET_ERROR(EIO)); } retry: dr = vdev_disk_dio_alloc(bio_count); if (zio && !(zio->io_flags & (ZIO_FLAG_IO_RETRY | ZIO_FLAG_TRYHARD))) bio_set_flags_failfast(bdev, &flags); dr->dr_zio = zio; /* * Since bio's can have up to BIO_MAX_PAGES=256 iovec's, each of which * is at least 512 bytes and at most PAGESIZE (typically 4K), one bio * can cover at least 128KB and at most 1MB. When the required number * of iovec's exceeds this, we are forced to break the IO in multiple * bio's and wait for them all to complete. This is likely if the * recordsize property is increased beyond 1MB. The default * bio_count=16 should typically accommodate the maximum-size zio of * 16MB. */ abd_offset = 0; bio_offset = io_offset; bio_size = io_size; for (int i = 0; i <= dr->dr_bio_count; i++) { /* Finished constructing bio's for given buffer */ if (bio_size <= 0) break; /* * If additional bio's are required, we have to retry, but * this should be rare - see the comment above. */ if (dr->dr_bio_count == i) { vdev_disk_dio_free(dr); bio_count *= 2; goto retry; } nr_vecs = vdev_bio_max_segs(zio, bio_size, abd_offset); dr->dr_bio[i] = vdev_bio_alloc(bdev, GFP_NOIO, nr_vecs); if (unlikely(dr->dr_bio[i] == NULL)) { vdev_disk_dio_free(dr); return (SET_ERROR(ENOMEM)); } /* Matching put called by vdev_disk_physio_completion */ vdev_disk_dio_get(dr); BIO_BI_SECTOR(dr->dr_bio[i]) = bio_offset >> 9; dr->dr_bio[i]->bi_end_io = vdev_disk_physio_completion; dr->dr_bio[i]->bi_private = dr; bio_set_op_attrs(dr->dr_bio[i], rw, flags); /* Remaining size is returned to become the new size */ bio_size = abd_bio_map_off(dr->dr_bio[i], zio->io_abd, bio_size, abd_offset); /* Advance in buffer and construct another bio if needed */ abd_offset += BIO_BI_SIZE(dr->dr_bio[i]); bio_offset += BIO_BI_SIZE(dr->dr_bio[i]); } /* Extra reference to protect dio_request during vdev_submit_bio */ vdev_disk_dio_get(dr); if (dr->dr_bio_count > 1) blk_start_plug(&plug); /* Submit all bio's associated with this dio */ for (int i = 0; i < dr->dr_bio_count; i++) { if (dr->dr_bio[i]) vdev_submit_bio(dr->dr_bio[i]); } if (dr->dr_bio_count > 1) blk_finish_plug(&plug); (void) vdev_disk_dio_put(dr); return (error); } BIO_END_IO_PROTO(vdev_disk_io_flush_completion, bio, error) { zio_t *zio = bio->bi_private; #ifdef HAVE_1ARG_BIO_END_IO_T zio->io_error = BIO_END_IO_ERROR(bio); #else zio->io_error = -error; #endif if (zio->io_error && (zio->io_error == EOPNOTSUPP)) zio->io_vd->vdev_nowritecache = B_TRUE; bio_put(bio); ASSERT3S(zio->io_error, >=, 0); if (zio->io_error) vdev_disk_error(zio); zio_interrupt(zio); } static int vdev_disk_io_flush(struct block_device *bdev, zio_t *zio) { struct request_queue *q; struct bio *bio; q = bdev_get_queue(bdev); if (!q) return (SET_ERROR(ENXIO)); bio = vdev_bio_alloc(bdev, GFP_NOIO, 0); if (unlikely(bio == NULL)) return (SET_ERROR(ENOMEM)); bio->bi_end_io = vdev_disk_io_flush_completion; bio->bi_private = zio; bio_set_flush(bio); vdev_submit_bio(bio); invalidate_bdev(bdev); return (0); } static int vdev_disk_io_trim(zio_t *zio) { vdev_t *v = zio->io_vd; vdev_disk_t *vd = v->vdev_tsd; #if defined(HAVE_BLKDEV_ISSUE_SECURE_ERASE) if (zio->io_trim_flags & ZIO_TRIM_SECURE) { return (-blkdev_issue_secure_erase(vd->vd_bdev, zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS)); } else { return (-blkdev_issue_discard(vd->vd_bdev, zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS)); } #elif defined(HAVE_BLKDEV_ISSUE_DISCARD) unsigned long trim_flags = 0; #if defined(BLKDEV_DISCARD_SECURE) if (zio->io_trim_flags & ZIO_TRIM_SECURE) trim_flags |= BLKDEV_DISCARD_SECURE; #endif return (-blkdev_issue_discard(vd->vd_bdev, zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS, trim_flags)); #else #error "Unsupported kernel" #endif } static void vdev_disk_io_start(zio_t *zio) { vdev_t *v = zio->io_vd; vdev_disk_t *vd = v->vdev_tsd; int rw, error; /* * If the vdev is closed, it's likely in the REMOVED or FAULTED state. * Nothing to be done here but return failure. */ if (vd == NULL) { zio->io_error = ENXIO; zio_interrupt(zio); return; } rw_enter(&vd->vd_lock, RW_READER); /* * If the vdev is closed, it's likely due to a failed reopen and is * in the UNAVAIL state. Nothing to be done here but return failure. */ if (vd->vd_bdev == NULL) { rw_exit(&vd->vd_lock); zio->io_error = ENXIO; zio_interrupt(zio); return; } switch (zio->io_type) { case ZIO_TYPE_IOCTL: if (!vdev_readable(v)) { rw_exit(&vd->vd_lock); zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return; } switch (zio->io_cmd) { case DKIOCFLUSHWRITECACHE: if (zfs_nocacheflush) break; if (v->vdev_nowritecache) { zio->io_error = SET_ERROR(ENOTSUP); break; } error = vdev_disk_io_flush(vd->vd_bdev, zio); if (error == 0) { rw_exit(&vd->vd_lock); return; } zio->io_error = error; break; default: zio->io_error = SET_ERROR(ENOTSUP); } rw_exit(&vd->vd_lock); zio_execute(zio); return; case ZIO_TYPE_WRITE: rw = WRITE; break; case ZIO_TYPE_READ: rw = READ; break; case ZIO_TYPE_TRIM: zio->io_error = vdev_disk_io_trim(zio); rw_exit(&vd->vd_lock); zio_interrupt(zio); return; default: rw_exit(&vd->vd_lock); zio->io_error = SET_ERROR(ENOTSUP); zio_interrupt(zio); return; } zio->io_target_timestamp = zio_handle_io_delay(zio); error = __vdev_disk_physio(vd->vd_bdev, zio, zio->io_size, zio->io_offset, rw, 0); rw_exit(&vd->vd_lock); if (error) { zio->io_error = error; zio_interrupt(zio); return; } } static void vdev_disk_io_done(zio_t *zio) { /* * If the device returned EIO, we revalidate the media. If it is * determined the media has changed this triggers the asynchronous * removal of the device from the configuration. */ if (zio->io_error == EIO) { vdev_t *v = zio->io_vd; vdev_disk_t *vd = v->vdev_tsd; - if (zfs_check_media_change(vd->vd_bdev)) { + if (!zfs_check_disk_status(vd->vd_bdev)) { invalidate_bdev(vd->vd_bdev); v->vdev_remove_wanted = B_TRUE; spa_async_request(zio->io_spa, SPA_ASYNC_REMOVE); } } } static void vdev_disk_hold(vdev_t *vd) { ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER)); /* We must have a pathname, and it must be absolute. */ if (vd->vdev_path == NULL || vd->vdev_path[0] != '/') return; /* * Only prefetch path and devid info if the device has * never been opened. */ if (vd->vdev_tsd != NULL) return; } static void vdev_disk_rele(vdev_t *vd) { ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER)); /* XXX: Implement me as a vnode rele for the device */ } vdev_ops_t vdev_disk_ops = { .vdev_op_init = NULL, .vdev_op_fini = NULL, .vdev_op_open = vdev_disk_open, .vdev_op_close = vdev_disk_close, .vdev_op_asize = vdev_default_asize, .vdev_op_min_asize = vdev_default_min_asize, .vdev_op_min_alloc = NULL, .vdev_op_io_start = vdev_disk_io_start, .vdev_op_io_done = vdev_disk_io_done, .vdev_op_state_change = NULL, .vdev_op_need_resilver = NULL, .vdev_op_hold = vdev_disk_hold, .vdev_op_rele = vdev_disk_rele, .vdev_op_remap = NULL, .vdev_op_xlate = vdev_default_xlate, .vdev_op_rebuild_asize = NULL, .vdev_op_metaslab_init = NULL, .vdev_op_config_generate = NULL, .vdev_op_nparity = NULL, .vdev_op_ndisks = NULL, .vdev_op_type = VDEV_TYPE_DISK, /* name of this vdev type */ - .vdev_op_leaf = B_TRUE /* leaf vdev */ + .vdev_op_leaf = B_TRUE, /* leaf vdev */ + .vdev_op_kobj_evt_post = vdev_disk_kobj_evt_post }; /* * The zfs_vdev_scheduler module option has been deprecated. Setting this * value no longer has any effect. It has not yet been entirely removed * to allow the module to be loaded if this option is specified in the * /etc/modprobe.d/zfs.conf file. The following warning will be logged. */ static int param_set_vdev_scheduler(const char *val, zfs_kernel_param_t *kp) { int error = param_set_charp(val, kp); if (error == 0) { printk(KERN_INFO "The 'zfs_vdev_scheduler' module option " "is not supported.\n"); } return (error); } static const char *zfs_vdev_scheduler = "unused"; module_param_call(zfs_vdev_scheduler, param_set_vdev_scheduler, param_get_charp, &zfs_vdev_scheduler, 0644); MODULE_PARM_DESC(zfs_vdev_scheduler, "I/O scheduler"); int param_set_min_auto_ashift(const char *buf, zfs_kernel_param_t *kp) { uint64_t val; int error; error = kstrtoull(buf, 0, &val); if (error < 0) return (SET_ERROR(error)); if (val < ASHIFT_MIN || val > zfs_vdev_max_auto_ashift) return (SET_ERROR(-EINVAL)); error = param_set_ulong(buf, kp); if (error < 0) return (SET_ERROR(error)); return (0); } int param_set_max_auto_ashift(const char *buf, zfs_kernel_param_t *kp) { uint64_t val; int error; error = kstrtoull(buf, 0, &val); if (error < 0) return (SET_ERROR(error)); if (val > ASHIFT_MAX || val < zfs_vdev_min_auto_ashift) return (SET_ERROR(-EINVAL)); error = param_set_ulong(buf, kp); if (error < 0) return (SET_ERROR(error)); return (0); } diff --git a/module/zfs/spa.c b/module/zfs/spa.c index b8e054a5f2a2..cc367745e486 100644 --- a/module/zfs/spa.c +++ b/module/zfs/spa.c @@ -1,10034 +1,10043 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2018, Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Toomas Soome * Copyright (c) 2016 Actifio, Inc. All rights reserved. * Copyright 2018 Joyent, Inc. * Copyright (c) 2017, 2019, Datto Inc. All rights reserved. * Copyright 2017 Joyent, Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2021, Colm Buckley */ /* * SPA: Storage Pool Allocator * * This file contains all the routines used when modifying on-disk SPA state. * This includes opening, importing, destroying, exporting a pool, and syncing a * pool. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #include #include #include #include #endif /* _KERNEL */ #include "zfs_prop.h" #include "zfs_comutil.h" /* * The interval, in seconds, at which failed configuration cache file writes * should be retried. */ int zfs_ccw_retry_interval = 300; typedef enum zti_modes { ZTI_MODE_FIXED, /* value is # of threads (min 1) */ ZTI_MODE_BATCH, /* cpu-intensive; value is ignored */ ZTI_MODE_SCALE, /* Taskqs scale with CPUs. */ ZTI_MODE_NULL, /* don't create a taskq */ ZTI_NMODES } zti_modes_t; #define ZTI_P(n, q) { ZTI_MODE_FIXED, (n), (q) } #define ZTI_PCT(n) { ZTI_MODE_ONLINE_PERCENT, (n), 1 } #define ZTI_BATCH { ZTI_MODE_BATCH, 0, 1 } #define ZTI_SCALE { ZTI_MODE_SCALE, 0, 1 } #define ZTI_NULL { ZTI_MODE_NULL, 0, 0 } #define ZTI_N(n) ZTI_P(n, 1) #define ZTI_ONE ZTI_N(1) typedef struct zio_taskq_info { zti_modes_t zti_mode; uint_t zti_value; uint_t zti_count; } zio_taskq_info_t; static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = { "iss", "iss_h", "int", "int_h" }; /* * This table defines the taskq settings for each ZFS I/O type. When * initializing a pool, we use this table to create an appropriately sized * taskq. Some operations are low volume and therefore have a small, static * number of threads assigned to their taskqs using the ZTI_N(#) or ZTI_ONE * macros. Other operations process a large amount of data; the ZTI_BATCH * macro causes us to create a taskq oriented for throughput. Some operations * are so high frequency and short-lived that the taskq itself can become a * point of lock contention. The ZTI_P(#, #) macro indicates that we need an * additional degree of parallelism specified by the number of threads per- * taskq and the number of taskqs; when dispatching an event in this case, the * particular taskq is chosen at random. ZTI_SCALE is similar to ZTI_BATCH, * but with number of taskqs also scaling with number of CPUs. * * The different taskq priorities are to handle the different contexts (issue * and interrupt) and then to reserve threads for ZIO_PRIORITY_NOW I/Os that * need to be handled with minimum delay. */ static 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 }, /* NULL */ { ZTI_N(8), ZTI_NULL, ZTI_SCALE, ZTI_NULL }, /* READ */ { ZTI_BATCH, ZTI_N(5), ZTI_SCALE, ZTI_N(5) }, /* WRITE */ { ZTI_SCALE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* FREE */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* CLAIM */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* IOCTL */ { ZTI_N(4), ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* TRIM */ }; static void spa_sync_version(void *arg, dmu_tx_t *tx); static void spa_sync_props(void *arg, dmu_tx_t *tx); static boolean_t spa_has_active_shared_spare(spa_t *spa); static int spa_load_impl(spa_t *spa, spa_import_type_t type, const char **ereport); static void spa_vdev_resilver_done(spa_t *spa); static uint_t zio_taskq_batch_pct = 80; /* 1 thread per cpu in pset */ static uint_t zio_taskq_batch_tpq; /* threads per taskq */ static const boolean_t zio_taskq_sysdc = B_TRUE; /* use SDC scheduling class */ static const uint_t zio_taskq_basedc = 80; /* base duty cycle */ static const boolean_t spa_create_process = B_TRUE; /* no process => no sysdc */ /* * Report any spa_load_verify errors found, but do not fail spa_load. * This is used by zdb to analyze non-idle pools. */ boolean_t spa_load_verify_dryrun = B_FALSE; /* * Allow read spacemaps in case of readonly import (spa_mode == SPA_MODE_READ). * This is used by zdb for spacemaps verification. */ boolean_t spa_mode_readable_spacemaps = B_FALSE; /* * This (illegal) pool name is used when temporarily importing a spa_t in order * to get the vdev stats associated with the imported devices. */ #define TRYIMPORT_NAME "$import" /* * For debugging purposes: print out vdev tree during pool import. */ static int spa_load_print_vdev_tree = B_FALSE; /* * A non-zero value for zfs_max_missing_tvds means that we allow importing * pools with missing top-level vdevs. This is strictly intended for advanced * pool recovery cases since missing data is almost inevitable. Pools with * missing devices can only be imported read-only for safety reasons, and their * fail-mode will be automatically set to "continue". * * With 1 missing vdev we should be able to import the pool and mount all * datasets. User data that was not modified after the missing device has been * added should be recoverable. This means that snapshots created prior to the * addition of that device should be completely intact. * * With 2 missing vdevs, some datasets may fail to mount since there are * dataset statistics that are stored as regular metadata. Some data might be * recoverable if those vdevs were added recently. * * With 3 or more missing vdevs, the pool is severely damaged and MOS entries * may be missing entirely. Chances of data recovery are very low. Note that * there are also risks of performing an inadvertent rewind as we might be * missing all the vdevs with the latest uberblocks. */ unsigned long zfs_max_missing_tvds = 0; /* * The parameters below are similar to zfs_max_missing_tvds but are only * intended for a preliminary open of the pool with an untrusted config which * might be incomplete or out-dated. * * We are more tolerant for pools opened from a cachefile since we could have * an out-dated cachefile where a device removal was not registered. * We could have set the limit arbitrarily high but in the case where devices * are really missing we would want to return the proper error codes; we chose * SPA_DVAS_PER_BP - 1 so that some copies of the MOS would still be available * and we get a chance to retrieve the trusted config. */ uint64_t zfs_max_missing_tvds_cachefile = SPA_DVAS_PER_BP - 1; /* * In the case where config was assembled by scanning device paths (/dev/dsks * by default) we are less tolerant since all the existing devices should have * been detected and we want spa_load to return the right error codes. */ uint64_t zfs_max_missing_tvds_scan = 0; /* * Debugging aid that pauses spa_sync() towards the end. */ static const boolean_t zfs_pause_spa_sync = B_FALSE; /* * Variables to indicate the livelist condense zthr func should wait at certain * points for the livelist to be removed - used to test condense/destroy races */ static int zfs_livelist_condense_zthr_pause = 0; static int zfs_livelist_condense_sync_pause = 0; /* * Variables to track whether or not condense cancellation has been * triggered in testing. */ static int zfs_livelist_condense_sync_cancel = 0; static int zfs_livelist_condense_zthr_cancel = 0; /* * Variable to track whether or not extra ALLOC blkptrs were added to a * livelist entry while it was being condensed (caused by the way we track * remapped blkptrs in dbuf_remap_impl) */ static int zfs_livelist_condense_new_alloc = 0; /* * ========================================================================== * SPA properties routines * ========================================================================== */ /* * Add a (source=src, propname=propval) list to an nvlist. */ static void spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, const char *strval, uint64_t intval, zprop_source_t src) { const char *propname = zpool_prop_to_name(prop); nvlist_t *propval; propval = fnvlist_alloc(); fnvlist_add_uint64(propval, ZPROP_SOURCE, src); if (strval != NULL) fnvlist_add_string(propval, ZPROP_VALUE, strval); else fnvlist_add_uint64(propval, ZPROP_VALUE, intval); fnvlist_add_nvlist(nvl, propname, propval); nvlist_free(propval); } /* * Get property values from the spa configuration. */ static void spa_prop_get_config(spa_t *spa, nvlist_t **nvp) { vdev_t *rvd = spa->spa_root_vdev; dsl_pool_t *pool = spa->spa_dsl_pool; uint64_t size, alloc, cap, version; const zprop_source_t src = ZPROP_SRC_NONE; spa_config_dirent_t *dp; metaslab_class_t *mc = spa_normal_class(spa); ASSERT(MUTEX_HELD(&spa->spa_props_lock)); if (rvd != NULL) { alloc = metaslab_class_get_alloc(mc); alloc += metaslab_class_get_alloc(spa_special_class(spa)); alloc += metaslab_class_get_alloc(spa_dedup_class(spa)); alloc += metaslab_class_get_alloc(spa_embedded_log_class(spa)); size = metaslab_class_get_space(mc); size += metaslab_class_get_space(spa_special_class(spa)); size += metaslab_class_get_space(spa_dedup_class(spa)); size += metaslab_class_get_space(spa_embedded_log_class(spa)); spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src); spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src); spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src); spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL, size - alloc, src); spa_prop_add_list(*nvp, ZPOOL_PROP_CHECKPOINT, NULL, spa->spa_checkpoint_info.sci_dspace, src); spa_prop_add_list(*nvp, ZPOOL_PROP_FRAGMENTATION, NULL, metaslab_class_fragmentation(mc), src); spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL, metaslab_class_expandable_space(mc), src); spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL, (spa_mode(spa) == SPA_MODE_READ), src); cap = (size == 0) ? 0 : (alloc * 100 / size); spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src); spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL, ddt_get_pool_dedup_ratio(spa), src); spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL, rvd->vdev_state, src); version = spa_version(spa); if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION)) { spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, ZPROP_SRC_DEFAULT); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, ZPROP_SRC_LOCAL); } spa_prop_add_list(*nvp, ZPOOL_PROP_LOAD_GUID, NULL, spa_load_guid(spa), src); } if (pool != NULL) { /* * The $FREE directory was introduced in SPA_VERSION_DEADLISTS, * when opening pools before this version freedir will be NULL. */ if (pool->dp_free_dir != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, dsl_dir_phys(pool->dp_free_dir)->dd_used_bytes, src); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, 0, src); } if (pool->dp_leak_dir != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL, dsl_dir_phys(pool->dp_leak_dir)->dd_used_bytes, src); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL, 0, src); } } spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src); if (spa->spa_comment != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_COMMENT, spa->spa_comment, 0, ZPROP_SRC_LOCAL); } if (spa->spa_compatibility != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_COMPATIBILITY, spa->spa_compatibility, 0, ZPROP_SRC_LOCAL); } if (spa->spa_root != NULL) spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root, 0, ZPROP_SRC_LOCAL); if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) { spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL, MIN(zfs_max_recordsize, SPA_MAXBLOCKSIZE), ZPROP_SRC_NONE); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL, SPA_OLD_MAXBLOCKSIZE, ZPROP_SRC_NONE); } if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) { spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL, DNODE_MAX_SIZE, ZPROP_SRC_NONE); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL, DNODE_MIN_SIZE, ZPROP_SRC_NONE); } if ((dp = list_head(&spa->spa_config_list)) != NULL) { if (dp->scd_path == NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, "none", 0, ZPROP_SRC_LOCAL); } else if (strcmp(dp->scd_path, spa_config_path) != 0) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, dp->scd_path, 0, ZPROP_SRC_LOCAL); } } } /* * Get zpool property values. */ int spa_prop_get(spa_t *spa, nvlist_t **nvp) { objset_t *mos = spa->spa_meta_objset; zap_cursor_t zc; zap_attribute_t za; dsl_pool_t *dp; int err; err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP); if (err) return (err); dp = spa_get_dsl(spa); dsl_pool_config_enter(dp, FTAG); mutex_enter(&spa->spa_props_lock); /* * Get properties from the spa config. */ spa_prop_get_config(spa, nvp); /* If no pool property object, no more prop to get. */ if (mos == NULL || spa->spa_pool_props_object == 0) goto out; /* * Get properties from the MOS pool property object. */ for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object); (err = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { uint64_t intval = 0; char *strval = NULL; zprop_source_t src = ZPROP_SRC_DEFAULT; zpool_prop_t prop; if ((prop = zpool_name_to_prop(za.za_name)) == ZPOOL_PROP_INVAL) continue; switch (za.za_integer_length) { case 8: /* integer property */ if (za.za_first_integer != zpool_prop_default_numeric(prop)) src = ZPROP_SRC_LOCAL; if (prop == ZPOOL_PROP_BOOTFS) { dsl_dataset_t *ds = NULL; err = dsl_dataset_hold_obj(dp, za.za_first_integer, FTAG, &ds); if (err != 0) break; strval = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); dsl_dataset_name(ds, strval); dsl_dataset_rele(ds, FTAG); } else { strval = NULL; intval = za.za_first_integer; } spa_prop_add_list(*nvp, prop, strval, intval, src); if (strval != NULL) kmem_free(strval, ZFS_MAX_DATASET_NAME_LEN); break; case 1: /* string property */ strval = kmem_alloc(za.za_num_integers, KM_SLEEP); err = zap_lookup(mos, spa->spa_pool_props_object, za.za_name, 1, za.za_num_integers, strval); if (err) { kmem_free(strval, za.za_num_integers); break; } spa_prop_add_list(*nvp, prop, strval, 0, src); kmem_free(strval, za.za_num_integers); break; default: break; } } zap_cursor_fini(&zc); out: mutex_exit(&spa->spa_props_lock); dsl_pool_config_exit(dp, FTAG); if (err && err != ENOENT) { nvlist_free(*nvp); *nvp = NULL; return (err); } return (0); } /* * Validate the given pool properties nvlist and modify the list * for the property values to be set. */ static int spa_prop_validate(spa_t *spa, nvlist_t *props) { nvpair_t *elem; int error = 0, reset_bootfs = 0; uint64_t objnum = 0; boolean_t has_feature = B_FALSE; elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { uint64_t intval; char *strval, *slash, *check, *fname; const char *propname = nvpair_name(elem); zpool_prop_t prop = zpool_name_to_prop(propname); switch (prop) { case ZPOOL_PROP_INVAL: if (!zpool_prop_feature(propname)) { error = SET_ERROR(EINVAL); break; } /* * Sanitize the input. */ if (nvpair_type(elem) != DATA_TYPE_UINT64) { error = SET_ERROR(EINVAL); break; } if (nvpair_value_uint64(elem, &intval) != 0) { error = SET_ERROR(EINVAL); break; } if (intval != 0) { error = SET_ERROR(EINVAL); break; } fname = strchr(propname, '@') + 1; if (zfeature_lookup_name(fname, NULL) != 0) { error = SET_ERROR(EINVAL); break; } has_feature = B_TRUE; break; case ZPOOL_PROP_VERSION: error = nvpair_value_uint64(elem, &intval); if (!error && (intval < spa_version(spa) || intval > SPA_VERSION_BEFORE_FEATURES || has_feature)) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_DELEGATION: case ZPOOL_PROP_AUTOREPLACE: case ZPOOL_PROP_LISTSNAPS: case ZPOOL_PROP_AUTOEXPAND: case ZPOOL_PROP_AUTOTRIM: error = nvpair_value_uint64(elem, &intval); if (!error && intval > 1) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_MULTIHOST: error = nvpair_value_uint64(elem, &intval); if (!error && intval > 1) error = SET_ERROR(EINVAL); if (!error) { uint32_t hostid = zone_get_hostid(NULL); if (hostid) spa->spa_hostid = hostid; else error = SET_ERROR(ENOTSUP); } break; case ZPOOL_PROP_BOOTFS: /* * If the pool version is less than SPA_VERSION_BOOTFS, * or the pool is still being created (version == 0), * the bootfs property cannot be set. */ if (spa_version(spa) < SPA_VERSION_BOOTFS) { error = SET_ERROR(ENOTSUP); break; } /* * Make sure the vdev config is bootable */ if (!vdev_is_bootable(spa->spa_root_vdev)) { error = SET_ERROR(ENOTSUP); break; } reset_bootfs = 1; error = nvpair_value_string(elem, &strval); if (!error) { objset_t *os; if (strval == NULL || strval[0] == '\0') { objnum = zpool_prop_default_numeric( ZPOOL_PROP_BOOTFS); break; } error = dmu_objset_hold(strval, FTAG, &os); if (error != 0) break; /* Must be ZPL. */ if (dmu_objset_type(os) != DMU_OST_ZFS) { error = SET_ERROR(ENOTSUP); } else { objnum = dmu_objset_id(os); } dmu_objset_rele(os, FTAG); } break; case ZPOOL_PROP_FAILUREMODE: error = nvpair_value_uint64(elem, &intval); if (!error && intval > ZIO_FAILURE_MODE_PANIC) error = SET_ERROR(EINVAL); /* * This is a special case which only occurs when * the pool has completely failed. This allows * the user to change the in-core failmode property * without syncing it out to disk (I/Os might * currently be blocked). We do this by returning * EIO to the caller (spa_prop_set) to trick it * into thinking we encountered a property validation * error. */ if (!error && spa_suspended(spa)) { spa->spa_failmode = intval; error = SET_ERROR(EIO); } break; case ZPOOL_PROP_CACHEFILE: if ((error = nvpair_value_string(elem, &strval)) != 0) break; if (strval[0] == '\0') break; if (strcmp(strval, "none") == 0) break; if (strval[0] != '/') { error = SET_ERROR(EINVAL); break; } slash = strrchr(strval, '/'); ASSERT(slash != NULL); if (slash[1] == '\0' || strcmp(slash, "/.") == 0 || strcmp(slash, "/..") == 0) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_COMMENT: if ((error = nvpair_value_string(elem, &strval)) != 0) break; for (check = strval; *check != '\0'; check++) { if (!isprint(*check)) { error = SET_ERROR(EINVAL); break; } } if (strlen(strval) > ZPROP_MAX_COMMENT) error = SET_ERROR(E2BIG); break; default: break; } if (error) break; } (void) nvlist_remove_all(props, zpool_prop_to_name(ZPOOL_PROP_DEDUPDITTO)); if (!error && reset_bootfs) { error = nvlist_remove(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING); if (!error) { error = nvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum); } } return (error); } void spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync) { char *cachefile; spa_config_dirent_t *dp; if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), &cachefile) != 0) return; dp = kmem_alloc(sizeof (spa_config_dirent_t), KM_SLEEP); if (cachefile[0] == '\0') dp->scd_path = spa_strdup(spa_config_path); else if (strcmp(cachefile, "none") == 0) dp->scd_path = NULL; else dp->scd_path = spa_strdup(cachefile); list_insert_head(&spa->spa_config_list, dp); if (need_sync) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } int spa_prop_set(spa_t *spa, nvlist_t *nvp) { int error; nvpair_t *elem = NULL; boolean_t need_sync = B_FALSE; if ((error = spa_prop_validate(spa, nvp)) != 0) return (error); while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) { zpool_prop_t prop = zpool_name_to_prop(nvpair_name(elem)); if (prop == ZPOOL_PROP_CACHEFILE || prop == ZPOOL_PROP_ALTROOT || prop == ZPOOL_PROP_READONLY) continue; if (prop == ZPOOL_PROP_VERSION || prop == ZPOOL_PROP_INVAL) { uint64_t ver = 0; if (prop == ZPOOL_PROP_VERSION) { VERIFY(nvpair_value_uint64(elem, &ver) == 0); } else { ASSERT(zpool_prop_feature(nvpair_name(elem))); ver = SPA_VERSION_FEATURES; need_sync = B_TRUE; } /* Save time if the version is already set. */ if (ver == spa_version(spa)) continue; /* * In addition to the pool directory object, we might * create the pool properties object, the features for * read object, the features for write object, or the * feature descriptions object. */ error = dsl_sync_task(spa->spa_name, NULL, spa_sync_version, &ver, 6, ZFS_SPACE_CHECK_RESERVED); if (error) return (error); continue; } need_sync = B_TRUE; break; } if (need_sync) { return (dsl_sync_task(spa->spa_name, NULL, spa_sync_props, nvp, 6, ZFS_SPACE_CHECK_RESERVED)); } return (0); } /* * If the bootfs property value is dsobj, clear it. */ void spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx) { if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) { VERIFY(zap_remove(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0); spa->spa_bootfs = 0; } } static int spa_change_guid_check(void *arg, dmu_tx_t *tx) { uint64_t *newguid __maybe_unused = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; vdev_t *rvd = spa->spa_root_vdev; uint64_t vdev_state; if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) { int error = (spa_has_checkpoint(spa)) ? ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT; return (SET_ERROR(error)); } spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); vdev_state = rvd->vdev_state; spa_config_exit(spa, SCL_STATE, FTAG); if (vdev_state != VDEV_STATE_HEALTHY) return (SET_ERROR(ENXIO)); ASSERT3U(spa_guid(spa), !=, *newguid); return (0); } static void spa_change_guid_sync(void *arg, dmu_tx_t *tx) { uint64_t *newguid = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; uint64_t oldguid; vdev_t *rvd = spa->spa_root_vdev; oldguid = spa_guid(spa); spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); rvd->vdev_guid = *newguid; rvd->vdev_guid_sum += (*newguid - oldguid); vdev_config_dirty(rvd); spa_config_exit(spa, SCL_STATE, FTAG); spa_history_log_internal(spa, "guid change", tx, "old=%llu new=%llu", (u_longlong_t)oldguid, (u_longlong_t)*newguid); } /* * Change the GUID for the pool. This is done so that we can later * re-import a pool built from a clone of our own vdevs. We will modify * the root vdev's guid, our own pool guid, and then mark all of our * vdevs dirty. Note that we must make sure that all our vdevs are * online when we do this, or else any vdevs that weren't present * would be orphaned from our pool. We are also going to issue a * sysevent to update any watchers. */ int spa_change_guid(spa_t *spa) { int error; uint64_t guid; mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); guid = spa_generate_guid(NULL); error = dsl_sync_task(spa->spa_name, spa_change_guid_check, spa_change_guid_sync, &guid, 5, ZFS_SPACE_CHECK_RESERVED); if (error == 0) { - spa_write_cachefile(spa, B_FALSE, B_TRUE); + /* + * Clear the kobj flag from all the vdevs to allow + * vdev_cache_process_kobj_evt() to post events to all the + * vdevs since GUID is updated. + */ + vdev_clear_kobj_evt(spa->spa_root_vdev); + for (int i = 0; i < spa->spa_l2cache.sav_count; i++) + vdev_clear_kobj_evt(spa->spa_l2cache.sav_vdevs[i]); + + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_TRUE); spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_REGUID); } mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * ========================================================================== * SPA state manipulation (open/create/destroy/import/export) * ========================================================================== */ static int spa_error_entry_compare(const void *a, const void *b) { const spa_error_entry_t *sa = (const spa_error_entry_t *)a; const spa_error_entry_t *sb = (const spa_error_entry_t *)b; int ret; ret = memcmp(&sa->se_bookmark, &sb->se_bookmark, sizeof (zbookmark_phys_t)); return (TREE_ISIGN(ret)); } /* * Utility function which retrieves copies of the current logs and * re-initializes them in the process. */ void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub) { ASSERT(MUTEX_HELD(&spa->spa_errlist_lock)); memcpy(last, &spa->spa_errlist_last, sizeof (avl_tree_t)); memcpy(scrub, &spa->spa_errlist_scrub, sizeof (avl_tree_t)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); } static void spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { 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 count = ztip->zti_count; spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t cpus, flags = TASKQ_DYNAMIC; boolean_t batch = B_FALSE; switch (mode) { case ZTI_MODE_FIXED: ASSERT3U(value, >, 0); break; case ZTI_MODE_BATCH: batch = B_TRUE; flags |= TASKQ_THREADS_CPU_PCT; value = MIN(zio_taskq_batch_pct, 100); break; case ZTI_MODE_SCALE: flags |= TASKQ_THREADS_CPU_PCT; /* * We want more taskqs to reduce lock contention, but we want * less for better request ordering and CPU utilization. */ cpus = MAX(1, boot_ncpus * zio_taskq_batch_pct / 100); if (zio_taskq_batch_tpq > 0) { count = MAX(1, (cpus + zio_taskq_batch_tpq / 2) / zio_taskq_batch_tpq); } else { /* * Prefer 6 threads per taskq, but no more taskqs * than threads in them on large systems. For 80%: * * taskq taskq total * cpus taskqs percent threads threads * ------- ------- ------- ------- ------- * 1 1 80% 1 1 * 2 1 80% 1 1 * 4 1 80% 3 3 * 8 2 40% 3 6 * 16 3 27% 4 12 * 32 5 16% 5 25 * 64 7 11% 7 49 * 128 10 8% 10 100 * 256 14 6% 15 210 */ count = 1 + cpus / 6; while (count * count > cpus) count--; } /* Limit each taskq within 100% to not trigger assertion. */ count = MAX(count, (zio_taskq_batch_pct + 99) / 100); value = (zio_taskq_batch_pct + count / 2) / count; break; case ZTI_MODE_NULL: tqs->stqs_count = 0; tqs->stqs_taskq = NULL; return; default: panic("unrecognized mode for %s_%s taskq (%u:%u) in " "spa_activate()", zio_type_name[t], zio_taskq_types[q], mode, value); break; } ASSERT3U(count, >, 0); tqs->stqs_count = count; tqs->stqs_taskq = kmem_alloc(count * sizeof (taskq_t *), KM_SLEEP); for (uint_t i = 0; i < count; i++) { taskq_t *tq; char name[32]; if (count > 1) (void) snprintf(name, sizeof (name), "%s_%s_%u", zio_type_name[t], zio_taskq_types[q], i); else (void) snprintf(name, sizeof (name), "%s_%s", zio_type_name[t], zio_taskq_types[q]); if (zio_taskq_sysdc && spa->spa_proc != &p0) { if (batch) flags |= TASKQ_DC_BATCH; (void) zio_taskq_basedc; tq = taskq_create_sysdc(name, value, 50, INT_MAX, spa->spa_proc, zio_taskq_basedc, flags); } else { pri_t pri = maxclsyspri; /* * The write issue taskq can be extremely CPU * intensive. Run it at slightly less important * priority than the other taskqs. * * Under Linux and FreeBSD this means incrementing * the priority value as opposed to platforms like * illumos where it should be decremented. * * On FreeBSD, if priorities divided by four (RQ_PPQ) * are equal then a difference between them is * insignificant. */ if (t == ZIO_TYPE_WRITE && q == ZIO_TASKQ_ISSUE) { #if defined(__linux__) pri++; #elif defined(__FreeBSD__) pri += 4; #else #error "unknown OS" #endif } tq = taskq_create_proc(name, value, pri, 50, INT_MAX, spa->spa_proc, flags); } tqs->stqs_taskq[i] = tq; } } static void spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; if (tqs->stqs_taskq == NULL) { ASSERT3U(tqs->stqs_count, ==, 0); return; } for (uint_t i = 0; i < tqs->stqs_count; i++) { ASSERT3P(tqs->stqs_taskq[i], !=, NULL); taskq_destroy(tqs->stqs_taskq[i]); } kmem_free(tqs->stqs_taskq, tqs->stqs_count * sizeof (taskq_t *)); tqs->stqs_taskq = NULL; } /* * Dispatch a task to the appropriate taskq for the ZFS I/O type and priority. * Note that a type may have multiple discrete taskqs to avoid lock contention * on the taskq itself. In that case we choose which taskq at random by using * the low bits of gethrtime(). */ void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } taskq_dispatch_ent(tq, func, arg, flags, ent); } /* * Same as spa_taskq_dispatch_ent() but block on the task until completion. */ void spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; taskqid_t id; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } id = taskq_dispatch(tq, func, arg, flags); if (id) taskq_wait_id(tq, id); } static void spa_create_zio_taskqs(spa_t *spa) { for (int t = 0; t < ZIO_TYPES; t++) { for (int q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_init(spa, t, q); } } } /* * Disabled until spa_thread() can be adapted for Linux. */ #undef HAVE_SPA_THREAD #if defined(_KERNEL) && defined(HAVE_SPA_THREAD) static void spa_thread(void *arg) { psetid_t zio_taskq_psrset_bind = PS_NONE; callb_cpr_t cprinfo; spa_t *spa = arg; user_t *pu = PTOU(curproc); CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr, spa->spa_name); ASSERT(curproc != &p0); (void) snprintf(pu->u_psargs, sizeof (pu->u_psargs), "zpool-%s", spa->spa_name); (void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm)); /* bind this thread to the requested psrset */ if (zio_taskq_psrset_bind != PS_NONE) { pool_lock(); mutex_enter(&cpu_lock); mutex_enter(&pidlock); mutex_enter(&curproc->p_lock); if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind, 0, NULL, NULL) == 0) { curthread->t_bind_pset = zio_taskq_psrset_bind; } else { cmn_err(CE_WARN, "Couldn't bind process for zfs pool \"%s\" to " "pset %d\n", spa->spa_name, zio_taskq_psrset_bind); } mutex_exit(&curproc->p_lock); mutex_exit(&pidlock); mutex_exit(&cpu_lock); pool_unlock(); } if (zio_taskq_sysdc) { sysdc_thread_enter(curthread, 100, 0); } spa->spa_proc = curproc; spa->spa_did = curthread->t_did; spa_create_zio_taskqs(spa); mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_CREATED); spa->spa_proc_state = SPA_PROC_ACTIVE; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_SAFE_BEGIN(&cprinfo); while (spa->spa_proc_state == SPA_PROC_ACTIVE) cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE); spa->spa_proc_state = SPA_PROC_GONE; spa->spa_proc = &p0; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_EXIT(&cprinfo); /* drops spa_proc_lock */ mutex_enter(&curproc->p_lock); lwp_exit(); } #endif /* * Activate an uninitialized pool. */ static void spa_activate(spa_t *spa, spa_mode_t mode) { ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED); spa->spa_state = POOL_STATE_ACTIVE; spa->spa_mode = mode; spa->spa_read_spacemaps = spa_mode_readable_spacemaps; spa->spa_normal_class = metaslab_class_create(spa, &zfs_metaslab_ops); spa->spa_log_class = metaslab_class_create(spa, &zfs_metaslab_ops); spa->spa_embedded_log_class = metaslab_class_create(spa, &zfs_metaslab_ops); spa->spa_special_class = metaslab_class_create(spa, &zfs_metaslab_ops); spa->spa_dedup_class = metaslab_class_create(spa, &zfs_metaslab_ops); /* Try to create a covering process */ mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_NONE); ASSERT(spa->spa_proc == &p0); spa->spa_did = 0; (void) spa_create_process; #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, NULL, 0) == 0) { spa->spa_proc_state = SPA_PROC_CREATED; while (spa->spa_proc_state == SPA_PROC_CREATED) { cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); ASSERT(spa->spa_proc != &p0); ASSERT(spa->spa_did != 0); } else { #ifdef _KERNEL cmn_err(CE_WARN, "Couldn't create process for zfs pool \"%s\"\n", spa->spa_name); #endif } } #endif /* HAVE_SPA_THREAD */ mutex_exit(&spa->spa_proc_lock); /* If we didn't create a process, we need to create our taskqs. */ if (spa->spa_proc == &p0) { spa_create_zio_taskqs(spa); } for (size_t i = 0; i < TXG_SIZE; i++) { spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); } list_create(&spa->spa_config_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_config_dirty_node)); list_create(&spa->spa_evicting_os_list, sizeof (objset_t), offsetof(objset_t, os_evicting_node)); list_create(&spa->spa_state_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_state_dirty_node)); txg_list_create(&spa->spa_vdev_txg_list, spa, offsetof(struct vdev, vdev_txg_node)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_healed, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); spa_activate_os(spa); spa_keystore_init(&spa->spa_keystore); /* * This taskq is used to perform zvol-minor-related tasks * asynchronously. This has several advantages, including easy * resolution of various deadlocks. * * The taskq must be single threaded to ensure tasks are always * processed in the order in which they were dispatched. * * A taskq per pool allows one to keep the pools independent. * This way if one pool is suspended, it will not impact another. * * The preferred location to dispatch a zvol minor task is a sync * task. In this context, there is easy access to the spa_t and minimal * error handling is required because the sync task must succeed. */ spa->spa_zvol_taskq = taskq_create("z_zvol", 1, defclsyspri, 1, INT_MAX, 0); /* * Taskq dedicated to prefetcher threads: this is used to prevent the * pool traverse code from monopolizing the global (and limited) * system_taskq by inappropriately scheduling long running tasks on it. */ spa->spa_prefetch_taskq = taskq_create("z_prefetch", 100, defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC | TASKQ_THREADS_CPU_PCT); /* * The taskq to upgrade datasets in this pool. Currently used by * feature SPA_FEATURE_USEROBJ_ACCOUNTING/SPA_FEATURE_PROJECT_QUOTA. */ spa->spa_upgrade_taskq = taskq_create("z_upgrade", 100, defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC | TASKQ_THREADS_CPU_PCT); } /* * Opposite of spa_activate(). */ static void spa_deactivate(spa_t *spa) { ASSERT(spa->spa_sync_on == B_FALSE); ASSERT(spa->spa_dsl_pool == NULL); ASSERT(spa->spa_root_vdev == NULL); ASSERT(spa->spa_async_zio_root == NULL); ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED); spa_evicting_os_wait(spa); if (spa->spa_zvol_taskq) { taskq_destroy(spa->spa_zvol_taskq); spa->spa_zvol_taskq = NULL; } if (spa->spa_prefetch_taskq) { taskq_destroy(spa->spa_prefetch_taskq); spa->spa_prefetch_taskq = NULL; } if (spa->spa_upgrade_taskq) { taskq_destroy(spa->spa_upgrade_taskq); spa->spa_upgrade_taskq = NULL; } txg_list_destroy(&spa->spa_vdev_txg_list); list_destroy(&spa->spa_config_dirty_list); list_destroy(&spa->spa_evicting_os_list); list_destroy(&spa->spa_state_dirty_list); taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid); for (int t = 0; t < ZIO_TYPES; t++) { for (int q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_fini(spa, t, q); } } for (size_t i = 0; i < TXG_SIZE; i++) { ASSERT3P(spa->spa_txg_zio[i], !=, NULL); VERIFY0(zio_wait(spa->spa_txg_zio[i])); spa->spa_txg_zio[i] = NULL; } metaslab_class_destroy(spa->spa_normal_class); spa->spa_normal_class = NULL; metaslab_class_destroy(spa->spa_log_class); spa->spa_log_class = NULL; metaslab_class_destroy(spa->spa_embedded_log_class); spa->spa_embedded_log_class = NULL; metaslab_class_destroy(spa->spa_special_class); spa->spa_special_class = NULL; metaslab_class_destroy(spa->spa_dedup_class); spa->spa_dedup_class = NULL; /* * If this was part of an import or the open otherwise failed, we may * still have errors left in the queues. Empty them just in case. */ spa_errlog_drain(spa); avl_destroy(&spa->spa_errlist_scrub); avl_destroy(&spa->spa_errlist_last); avl_destroy(&spa->spa_errlist_healed); spa_keystore_fini(&spa->spa_keystore); spa->spa_state = POOL_STATE_UNINITIALIZED; mutex_enter(&spa->spa_proc_lock); if (spa->spa_proc_state != SPA_PROC_NONE) { ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); spa->spa_proc_state = SPA_PROC_DEACTIVATE; cv_broadcast(&spa->spa_proc_cv); while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) { ASSERT(spa->spa_proc != &p0); cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_GONE); spa->spa_proc_state = SPA_PROC_NONE; } ASSERT(spa->spa_proc == &p0); mutex_exit(&spa->spa_proc_lock); /* * We want to make sure spa_thread() has actually exited the ZFS * module, so that the module can't be unloaded out from underneath * it. */ if (spa->spa_did != 0) { thread_join(spa->spa_did); spa->spa_did = 0; } spa_deactivate_os(spa); } /* * Verify a pool configuration, and construct the vdev tree appropriately. This * will create all the necessary vdevs in the appropriate layout, with each vdev * in the CLOSED state. This will prep the pool before open/creation/import. * All vdev validation is done by the vdev_alloc() routine. */ int spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int atype) { nvlist_t **child; uint_t children; int error; if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0) return (error); if ((*vdp)->vdev_ops->vdev_op_leaf) return (0); error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children); if (error == ENOENT) return (0); if (error) { vdev_free(*vdp); *vdp = NULL; return (SET_ERROR(EINVAL)); } for (int c = 0; c < children; c++) { vdev_t *vd; if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c, atype)) != 0) { vdev_free(*vdp); *vdp = NULL; return (error); } } ASSERT(*vdp != NULL); return (0); } static boolean_t spa_should_flush_logs_on_unload(spa_t *spa) { if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) return (B_FALSE); if (!spa_writeable(spa)) return (B_FALSE); if (!spa->spa_sync_on) return (B_FALSE); if (spa_state(spa) != POOL_STATE_EXPORTED) return (B_FALSE); if (zfs_keep_log_spacemaps_at_export) return (B_FALSE); return (B_TRUE); } /* * Opens a transaction that will set the flag that will instruct * spa_sync to attempt to flush all the metaslabs for that txg. */ static void spa_unload_log_sm_flush_all(spa_t *spa) { dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); ASSERT3U(spa->spa_log_flushall_txg, ==, 0); spa->spa_log_flushall_txg = dmu_tx_get_txg(tx); dmu_tx_commit(tx); txg_wait_synced(spa_get_dsl(spa), spa->spa_log_flushall_txg); } static void spa_unload_log_sm_metadata(spa_t *spa) { void *cookie = NULL; spa_log_sm_t *sls; while ((sls = avl_destroy_nodes(&spa->spa_sm_logs_by_txg, &cookie)) != NULL) { VERIFY0(sls->sls_mscount); kmem_free(sls, sizeof (spa_log_sm_t)); } for (log_summary_entry_t *e = list_head(&spa->spa_log_summary); e != NULL; e = list_head(&spa->spa_log_summary)) { VERIFY0(e->lse_mscount); list_remove(&spa->spa_log_summary, e); kmem_free(e, sizeof (log_summary_entry_t)); } spa->spa_unflushed_stats.sus_nblocks = 0; spa->spa_unflushed_stats.sus_memused = 0; spa->spa_unflushed_stats.sus_blocklimit = 0; } static void spa_destroy_aux_threads(spa_t *spa) { if (spa->spa_condense_zthr != NULL) { zthr_destroy(spa->spa_condense_zthr); spa->spa_condense_zthr = NULL; } if (spa->spa_checkpoint_discard_zthr != NULL) { zthr_destroy(spa->spa_checkpoint_discard_zthr); spa->spa_checkpoint_discard_zthr = NULL; } if (spa->spa_livelist_delete_zthr != NULL) { zthr_destroy(spa->spa_livelist_delete_zthr); spa->spa_livelist_delete_zthr = NULL; } if (spa->spa_livelist_condense_zthr != NULL) { zthr_destroy(spa->spa_livelist_condense_zthr); spa->spa_livelist_condense_zthr = NULL; } } /* * Opposite of spa_load(). */ static void spa_unload(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_state(spa) != POOL_STATE_UNINITIALIZED); spa_import_progress_remove(spa_guid(spa)); spa_load_note(spa, "UNLOADING"); spa_wake_waiters(spa); /* * If we have set the spa_final_txg, we have already performed the * tasks below in spa_export_common(). We should not redo it here since * we delay the final TXGs beyond what spa_final_txg is set at. */ if (spa->spa_final_txg == UINT64_MAX) { /* * If the log space map feature is enabled and the pool is * getting exported (but not destroyed), we want to spend some * time flushing as many metaslabs as we can in an attempt to * destroy log space maps and save import time. */ if (spa_should_flush_logs_on_unload(spa)) spa_unload_log_sm_flush_all(spa); /* * Stop async tasks. */ spa_async_suspend(spa); if (spa->spa_root_vdev) { vdev_t *root_vdev = spa->spa_root_vdev; vdev_initialize_stop_all(root_vdev, VDEV_INITIALIZE_ACTIVE); vdev_trim_stop_all(root_vdev, VDEV_TRIM_ACTIVE); vdev_autotrim_stop_all(spa); vdev_rebuild_stop_all(spa); } } /* * Stop syncing. */ if (spa->spa_sync_on) { txg_sync_stop(spa->spa_dsl_pool); spa->spa_sync_on = B_FALSE; } /* * This ensures that there is no async metaslab prefetching * while we attempt to unload the spa. */ if (spa->spa_root_vdev != NULL) { for (int c = 0; c < spa->spa_root_vdev->vdev_children; c++) { vdev_t *vc = spa->spa_root_vdev->vdev_child[c]; if (vc->vdev_mg != NULL) taskq_wait(vc->vdev_mg->mg_taskq); } } if (spa->spa_mmp.mmp_thread) mmp_thread_stop(spa); /* * Wait for any outstanding async I/O to complete. */ if (spa->spa_async_zio_root != NULL) { for (int i = 0; i < max_ncpus; i++) (void) zio_wait(spa->spa_async_zio_root[i]); kmem_free(spa->spa_async_zio_root, max_ncpus * sizeof (void *)); spa->spa_async_zio_root = NULL; } if (spa->spa_vdev_removal != NULL) { spa_vdev_removal_destroy(spa->spa_vdev_removal); spa->spa_vdev_removal = NULL; } spa_destroy_aux_threads(spa); spa_condense_fini(spa); bpobj_close(&spa->spa_deferred_bpobj); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); /* * Close all vdevs. */ if (spa->spa_root_vdev) vdev_free(spa->spa_root_vdev); ASSERT(spa->spa_root_vdev == NULL); /* * Close the dsl pool. */ if (spa->spa_dsl_pool) { dsl_pool_close(spa->spa_dsl_pool); spa->spa_dsl_pool = NULL; spa->spa_meta_objset = NULL; } ddt_unload(spa); spa_unload_log_sm_metadata(spa); /* * Drop and purge level 2 cache */ spa_l2cache_drop(spa); for (int i = 0; i < spa->spa_spares.sav_count; i++) vdev_free(spa->spa_spares.sav_vdevs[i]); if (spa->spa_spares.sav_vdevs) { kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); spa->spa_spares.sav_vdevs = NULL; } if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; } spa->spa_spares.sav_count = 0; for (int i = 0; i < spa->spa_l2cache.sav_count; i++) { vdev_clear_stats(spa->spa_l2cache.sav_vdevs[i]); vdev_free(spa->spa_l2cache.sav_vdevs[i]); } if (spa->spa_l2cache.sav_vdevs) { kmem_free(spa->spa_l2cache.sav_vdevs, spa->spa_l2cache.sav_count * sizeof (void *)); spa->spa_l2cache.sav_vdevs = NULL; } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; } spa->spa_l2cache.sav_count = 0; spa->spa_async_suspended = 0; spa->spa_indirect_vdevs_loaded = B_FALSE; if (spa->spa_comment != NULL) { spa_strfree(spa->spa_comment); spa->spa_comment = NULL; } if (spa->spa_compatibility != NULL) { spa_strfree(spa->spa_compatibility); spa->spa_compatibility = NULL; } spa_config_exit(spa, SCL_ALL, spa); } /* * Load (or re-load) the current list of vdevs describing the active spares for * this pool. When this is called, we have some form of basic information in * 'spa_spares.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. */ void spa_load_spares(spa_t *spa) { nvlist_t **spares; uint_t nspares; int i; vdev_t *vd, *tvd; #ifndef _KERNEL /* * zdb opens both the current state of the pool and the * checkpointed state (if present), with a different spa_t. * * As spare vdevs are shared among open pools, we skip loading * them when we load the checkpointed state of the pool. */ if (!spa_writeable(spa)) return; #endif ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * First, close and free any existing spare vdevs. */ for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; /* Undo the call to spa_activate() below */ if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL && tvd->vdev_isspare) spa_spare_remove(tvd); vdev_close(vd); vdev_free(vd); } if (spa->spa_spares.sav_vdevs) kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); if (spa->spa_spares.sav_config == NULL) nspares = 0; else VERIFY0(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares)); spa->spa_spares.sav_count = (int)nspares; spa->spa_spares.sav_vdevs = NULL; if (nspares == 0) return; /* * Construct the array of vdevs, opening them to get status in the * process. For each spare, there is potentially two different vdev_t * structures associated with it: one in the list of spares (used only * for basic validation purposes) and one in the active vdev * configuration (if it's spared in). During this phase we open and * validate each vdev on the spare list. If the vdev also exists in the * active configuration, then we also mark this vdev as an active spare. */ spa->spa_spares.sav_vdevs = kmem_zalloc(nspares * sizeof (void *), KM_SLEEP); for (i = 0; i < spa->spa_spares.sav_count; i++) { VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0, VDEV_ALLOC_SPARE) == 0); ASSERT(vd != NULL); spa->spa_spares.sav_vdevs[i] = vd; if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL) { if (!tvd->vdev_isspare) spa_spare_add(tvd); /* * We only mark the spare active if we were successfully * able to load the vdev. Otherwise, importing a pool * with a bad active spare would result in strange * behavior, because multiple pool would think the spare * is actively in use. * * There is a vulnerability here to an equally bizarre * circumstance, where a dead active spare is later * brought back to life (onlined or otherwise). Given * the rarity of this scenario, and the extra complexity * it adds, we ignore the possibility. */ if (!vdev_is_dead(tvd)) spa_spare_activate(tvd); } vd->vdev_top = vd; vd->vdev_aux = &spa->spa_spares; if (vdev_open(vd) != 0) continue; if (vdev_validate_aux(vd) == 0) spa_spare_add(vd); } /* * Recompute the stashed list of spares, with status information * this time. */ fnvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES); spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *), KM_SLEEP); for (i = 0; i < spa->spa_spares.sav_count; i++) spares[i] = vdev_config_generate(spa, spa->spa_spares.sav_vdevs[i], B_TRUE, VDEV_CONFIG_SPARE); fnvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares, spa->spa_spares.sav_count); for (i = 0; i < spa->spa_spares.sav_count; i++) nvlist_free(spares[i]); kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *)); } /* * Load (or re-load) the current list of vdevs describing the active l2cache for * this pool. When this is called, we have some form of basic information in * 'spa_l2cache.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. * Devices which are already active have their details maintained, and are * not re-opened. */ void spa_load_l2cache(spa_t *spa) { nvlist_t **l2cache = NULL; uint_t nl2cache; int i, j, oldnvdevs; uint64_t guid; vdev_t *vd, **oldvdevs, **newvdevs; spa_aux_vdev_t *sav = &spa->spa_l2cache; #ifndef _KERNEL /* * zdb opens both the current state of the pool and the * checkpointed state (if present), with a different spa_t. * * As L2 caches are part of the ARC which is shared among open * pools, we skip loading them when we load the checkpointed * state of the pool. */ if (!spa_writeable(spa)) return; #endif ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); oldvdevs = sav->sav_vdevs; oldnvdevs = sav->sav_count; sav->sav_vdevs = NULL; sav->sav_count = 0; if (sav->sav_config == NULL) { nl2cache = 0; newvdevs = NULL; goto out; } VERIFY0(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache)); newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_SLEEP); /* * Process new nvlist of vdevs. */ for (i = 0; i < nl2cache; i++) { guid = fnvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID); newvdevs[i] = NULL; for (j = 0; j < oldnvdevs; j++) { vd = oldvdevs[j]; if (vd != NULL && guid == vd->vdev_guid) { /* * Retain previous vdev for add/remove ops. */ newvdevs[i] = vd; oldvdevs[j] = NULL; break; } } if (newvdevs[i] == NULL) { /* * Create new vdev */ VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0, VDEV_ALLOC_L2CACHE) == 0); ASSERT(vd != NULL); newvdevs[i] = vd; /* * Commit this vdev as an l2cache device, * even if it fails to open. */ spa_l2cache_add(vd); vd->vdev_top = vd; vd->vdev_aux = sav; spa_l2cache_activate(vd); if (vdev_open(vd) != 0) continue; (void) vdev_validate_aux(vd); if (!vdev_is_dead(vd)) l2arc_add_vdev(spa, vd); /* * Upon cache device addition to a pool or pool * creation with a cache device or if the header * of the device is invalid we issue an async * TRIM command for the whole device which will * execute if l2arc_trim_ahead > 0. */ spa_async_request(spa, SPA_ASYNC_L2CACHE_TRIM); } } sav->sav_vdevs = newvdevs; sav->sav_count = (int)nl2cache; /* * Recompute the stashed list of l2cache devices, with status * information this time. */ fnvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE); if (sav->sav_count > 0) l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_SLEEP); for (i = 0; i < sav->sav_count; i++) l2cache[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE); fnvlist_add_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache, sav->sav_count); out: /* * Purge vdevs that were dropped */ for (i = 0; i < oldnvdevs; i++) { uint64_t pool; vd = oldvdevs[i]; if (vd != NULL) { ASSERT(vd->vdev_isl2cache); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); vdev_clear_stats(vd); vdev_free(vd); } } if (oldvdevs) kmem_free(oldvdevs, oldnvdevs * sizeof (void *)); for (i = 0; i < sav->sav_count; i++) nvlist_free(l2cache[i]); if (sav->sav_count) kmem_free(l2cache, sav->sav_count * sizeof (void *)); } static int load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value) { dmu_buf_t *db; char *packed = NULL; size_t nvsize = 0; int error; *value = NULL; error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db); if (error) return (error); nvsize = *(uint64_t *)db->db_data; dmu_buf_rele(db, FTAG); packed = vmem_alloc(nvsize, KM_SLEEP); error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed, DMU_READ_PREFETCH); if (error == 0) error = nvlist_unpack(packed, nvsize, value, 0); vmem_free(packed, nvsize); return (error); } /* * Concrete top-level vdevs that are not missing and are not logs. At every * spa_sync we write new uberblocks to at least SPA_SYNC_MIN_VDEVS core tvds. */ static uint64_t spa_healthy_core_tvds(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; uint64_t tvds = 0; for (uint64_t i = 0; i < rvd->vdev_children; i++) { vdev_t *vd = rvd->vdev_child[i]; if (vd->vdev_islog) continue; if (vdev_is_concrete(vd) && !vdev_is_dead(vd)) tvds++; } return (tvds); } /* * Checks to see if the given vdev could not be opened, in which case we post a * sysevent to notify the autoreplace code that the device has been removed. */ static void spa_check_removed(vdev_t *vd) { for (uint64_t c = 0; c < vd->vdev_children; c++) spa_check_removed(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd) && vdev_is_concrete(vd)) { zfs_post_autoreplace(vd->vdev_spa, vd); spa_event_notify(vd->vdev_spa, vd, NULL, ESC_ZFS_VDEV_CHECK); } } static int spa_check_for_missing_logs(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; /* * If we're doing a normal import, then build up any additional * diagnostic information about missing log devices. * We'll pass this up to the user for further processing. */ if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) { nvlist_t **child, *nv; uint64_t idx = 0; child = kmem_alloc(rvd->vdev_children * sizeof (nvlist_t *), KM_SLEEP); nv = fnvlist_alloc(); for (uint64_t c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; /* * We consider a device as missing only if it failed * to open (i.e. offline or faulted is not considered * as missing). */ if (tvd->vdev_islog && tvd->vdev_state == VDEV_STATE_CANT_OPEN) { child[idx++] = vdev_config_generate(spa, tvd, B_FALSE, VDEV_CONFIG_MISSING); } } if (idx > 0) { fnvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, (const nvlist_t * const *)child, idx); fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_MISSING_DEVICES, nv); for (uint64_t i = 0; i < idx; i++) nvlist_free(child[i]); } nvlist_free(nv); kmem_free(child, rvd->vdev_children * sizeof (char **)); if (idx > 0) { spa_load_failed(spa, "some log devices are missing"); vdev_dbgmsg_print_tree(rvd, 2); return (SET_ERROR(ENXIO)); } } else { for (uint64_t c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; if (tvd->vdev_islog && tvd->vdev_state == VDEV_STATE_CANT_OPEN) { spa_set_log_state(spa, SPA_LOG_CLEAR); spa_load_note(spa, "some log devices are " "missing, ZIL is dropped."); vdev_dbgmsg_print_tree(rvd, 2); break; } } } return (0); } /* * Check for missing log devices */ static boolean_t spa_check_logs(spa_t *spa) { boolean_t rv = B_FALSE; dsl_pool_t *dp = spa_get_dsl(spa); switch (spa->spa_log_state) { default: break; case SPA_LOG_MISSING: /* need to recheck in case slog has been restored */ case SPA_LOG_UNKNOWN: rv = (dmu_objset_find_dp(dp, dp->dp_root_dir_obj, zil_check_log_chain, NULL, DS_FIND_CHILDREN) != 0); if (rv) spa_set_log_state(spa, SPA_LOG_MISSING); break; } return (rv); } /* * Passivate any log vdevs (note, does not apply to embedded log metaslabs). */ static boolean_t spa_passivate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; boolean_t slog_found = B_FALSE; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; if (tvd->vdev_islog) { ASSERT3P(tvd->vdev_log_mg, ==, NULL); metaslab_group_passivate(tvd->vdev_mg); slog_found = B_TRUE; } } return (slog_found); } /* * Activate any log vdevs (note, does not apply to embedded log metaslabs). */ static void spa_activate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; if (tvd->vdev_islog) { ASSERT3P(tvd->vdev_log_mg, ==, NULL); metaslab_group_activate(tvd->vdev_mg); } } } int spa_reset_logs(spa_t *spa) { int error; error = dmu_objset_find(spa_name(spa), zil_reset, NULL, DS_FIND_CHILDREN); if (error == 0) { /* * We successfully offlined the log device, sync out the * current txg so that the "stubby" block can be removed * by zil_sync(). */ txg_wait_synced(spa->spa_dsl_pool, 0); } return (error); } static void spa_aux_check_removed(spa_aux_vdev_t *sav) { for (int i = 0; i < sav->sav_count; i++) spa_check_removed(sav->sav_vdevs[i]); } void spa_claim_notify(zio_t *zio) { spa_t *spa = zio->io_spa; if (zio->io_error) return; mutex_enter(&spa->spa_props_lock); /* any mutex will do */ if (spa->spa_claim_max_txg < zio->io_bp->blk_birth) spa->spa_claim_max_txg = zio->io_bp->blk_birth; mutex_exit(&spa->spa_props_lock); } typedef struct spa_load_error { boolean_t sle_verify_data; uint64_t sle_meta_count; uint64_t sle_data_count; } spa_load_error_t; static void spa_load_verify_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; spa_load_error_t *sle = zio->io_private; dmu_object_type_t type = BP_GET_TYPE(bp); int error = zio->io_error; spa_t *spa = zio->io_spa; abd_free(zio->io_abd); if (error) { if ((BP_GET_LEVEL(bp) != 0 || DMU_OT_IS_METADATA(type)) && type != DMU_OT_INTENT_LOG) atomic_inc_64(&sle->sle_meta_count); else atomic_inc_64(&sle->sle_data_count); } mutex_enter(&spa->spa_scrub_lock); spa->spa_load_verify_bytes -= BP_GET_PSIZE(bp); cv_broadcast(&spa->spa_scrub_io_cv); mutex_exit(&spa->spa_scrub_lock); } /* * Maximum number of inflight bytes is the log2 fraction of the arc size. * By default, we set it to 1/16th of the arc. */ static uint_t spa_load_verify_shift = 4; static int spa_load_verify_metadata = B_TRUE; static int spa_load_verify_data = B_TRUE; static int spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg) { zio_t *rio = arg; spa_load_error_t *sle = rio->io_private; (void) zilog, (void) dnp; /* * Note: normally this routine will not be called if * spa_load_verify_metadata is not set. However, it may be useful * to manually set the flag after the traversal has begun. */ if (!spa_load_verify_metadata) return (0); /* * Sanity check the block pointer in order to detect obvious damage * before using the contents in subsequent checks or in zio_read(). * When damaged consider it to be a metadata error since we cannot * trust the BP_GET_TYPE and BP_GET_LEVEL values. */ if (!zfs_blkptr_verify(spa, bp, B_FALSE, BLK_VERIFY_LOG)) { atomic_inc_64(&sle->sle_meta_count); return (0); } if (zb->zb_level == ZB_DNODE_LEVEL || BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp) || BP_IS_REDACTED(bp)) return (0); if (!BP_IS_METADATA(bp) && (!spa_load_verify_data || !sle->sle_verify_data)) return (0); uint64_t maxinflight_bytes = arc_target_bytes() >> spa_load_verify_shift; size_t size = BP_GET_PSIZE(bp); mutex_enter(&spa->spa_scrub_lock); while (spa->spa_load_verify_bytes >= maxinflight_bytes) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_load_verify_bytes += size; mutex_exit(&spa->spa_scrub_lock); zio_nowait(zio_read(rio, spa, bp, abd_alloc_for_io(size, B_FALSE), size, spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB, ZIO_FLAG_SPECULATIVE | ZIO_FLAG_CANFAIL | ZIO_FLAG_SCRUB | ZIO_FLAG_RAW, zb)); return (0); } static int verify_dataset_name_len(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg) { (void) dp, (void) arg; if (dsl_dataset_namelen(ds) >= ZFS_MAX_DATASET_NAME_LEN) return (SET_ERROR(ENAMETOOLONG)); return (0); } static int spa_load_verify(spa_t *spa) { zio_t *rio; spa_load_error_t sle = { 0 }; zpool_load_policy_t policy; boolean_t verify_ok = B_FALSE; int error = 0; zpool_get_load_policy(spa->spa_config, &policy); if (policy.zlp_rewind & ZPOOL_NEVER_REWIND || policy.zlp_maxmeta == UINT64_MAX) return (0); dsl_pool_config_enter(spa->spa_dsl_pool, FTAG); error = dmu_objset_find_dp(spa->spa_dsl_pool, spa->spa_dsl_pool->dp_root_dir_obj, verify_dataset_name_len, NULL, DS_FIND_CHILDREN); dsl_pool_config_exit(spa->spa_dsl_pool, FTAG); if (error != 0) return (error); /* * Verify data only if we are rewinding or error limit was set. * Otherwise nothing except dbgmsg care about it to waste time. */ sle.sle_verify_data = (policy.zlp_rewind & ZPOOL_REWIND_MASK) || (policy.zlp_maxdata < UINT64_MAX); rio = zio_root(spa, NULL, &sle, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE); if (spa_load_verify_metadata) { if (spa->spa_extreme_rewind) { spa_load_note(spa, "performing a complete scan of the " "pool since extreme rewind is on. This may take " "a very long time.\n (spa_load_verify_data=%u, " "spa_load_verify_metadata=%u)", spa_load_verify_data, spa_load_verify_metadata); } error = traverse_pool(spa, spa->spa_verify_min_txg, TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA | TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio); } (void) zio_wait(rio); ASSERT0(spa->spa_load_verify_bytes); spa->spa_load_meta_errors = sle.sle_meta_count; spa->spa_load_data_errors = sle.sle_data_count; if (sle.sle_meta_count != 0 || sle.sle_data_count != 0) { spa_load_note(spa, "spa_load_verify found %llu metadata errors " "and %llu data errors", (u_longlong_t)sle.sle_meta_count, (u_longlong_t)sle.sle_data_count); } if (spa_load_verify_dryrun || (!error && sle.sle_meta_count <= policy.zlp_maxmeta && sle.sle_data_count <= policy.zlp_maxdata)) { int64_t loss = 0; verify_ok = B_TRUE; spa->spa_load_txg = spa->spa_uberblock.ub_txg; spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp; loss = spa->spa_last_ubsync_txg_ts - spa->spa_load_txg_ts; fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_TIME, spa->spa_load_txg_ts); fnvlist_add_int64(spa->spa_load_info, ZPOOL_CONFIG_REWIND_TIME, loss); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_META_ERRORS, sle.sle_meta_count); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_DATA_ERRORS, sle.sle_data_count); } else { spa->spa_load_max_txg = spa->spa_uberblock.ub_txg; } if (spa_load_verify_dryrun) return (0); if (error) { if (error != ENXIO && error != EIO) error = SET_ERROR(EIO); return (error); } return (verify_ok ? 0 : EIO); } /* * Find a value in the pool props object. */ static void spa_prop_find(spa_t *spa, zpool_prop_t prop, uint64_t *val) { (void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(prop), sizeof (uint64_t), 1, val); } /* * Find a value in the pool directory object. */ static int spa_dir_prop(spa_t *spa, const char *name, uint64_t *val, boolean_t log_enoent) { int error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, name, sizeof (uint64_t), 1, val); if (error != 0 && (error != ENOENT || log_enoent)) { spa_load_failed(spa, "couldn't get '%s' value in MOS directory " "[error=%d]", name, error); } return (error); } static int spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err) { vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux); return (SET_ERROR(err)); } boolean_t spa_livelist_delete_check(spa_t *spa) { return (spa->spa_livelists_to_delete != 0); } static boolean_t spa_livelist_delete_cb_check(void *arg, zthr_t *z) { (void) z; spa_t *spa = arg; return (spa_livelist_delete_check(spa)); } static int delete_blkptr_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { spa_t *spa = arg; zio_free(spa, tx->tx_txg, bp); dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD, -bp_get_dsize_sync(spa, bp), -BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx); return (0); } static int dsl_get_next_livelist_obj(objset_t *os, uint64_t zap_obj, uint64_t *llp) { int err; zap_cursor_t zc; zap_attribute_t za; zap_cursor_init(&zc, os, zap_obj); err = zap_cursor_retrieve(&zc, &za); zap_cursor_fini(&zc); if (err == 0) *llp = za.za_first_integer; return (err); } /* * Components of livelist deletion that must be performed in syncing * context: freeing block pointers and updating the pool-wide data * structures to indicate how much work is left to do */ typedef struct sublist_delete_arg { spa_t *spa; dsl_deadlist_t *ll; uint64_t key; bplist_t *to_free; } sublist_delete_arg_t; static void sublist_delete_sync(void *arg, dmu_tx_t *tx) { sublist_delete_arg_t *sda = arg; spa_t *spa = sda->spa; dsl_deadlist_t *ll = sda->ll; uint64_t key = sda->key; bplist_t *to_free = sda->to_free; bplist_iterate(to_free, delete_blkptr_cb, spa, tx); dsl_deadlist_remove_entry(ll, key, tx); } typedef struct livelist_delete_arg { spa_t *spa; uint64_t ll_obj; uint64_t zap_obj; } livelist_delete_arg_t; static void livelist_delete_sync(void *arg, dmu_tx_t *tx) { livelist_delete_arg_t *lda = arg; spa_t *spa = lda->spa; uint64_t ll_obj = lda->ll_obj; uint64_t zap_obj = lda->zap_obj; objset_t *mos = spa->spa_meta_objset; uint64_t count; /* free the livelist and decrement the feature count */ VERIFY0(zap_remove_int(mos, zap_obj, ll_obj, tx)); dsl_deadlist_free(mos, ll_obj, tx); spa_feature_decr(spa, SPA_FEATURE_LIVELIST, tx); VERIFY0(zap_count(mos, zap_obj, &count)); if (count == 0) { /* no more livelists to delete */ VERIFY0(zap_remove(mos, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DELETED_CLONES, tx)); VERIFY0(zap_destroy(mos, zap_obj, tx)); spa->spa_livelists_to_delete = 0; spa_notify_waiters(spa); } } /* * Load in the value for the livelist to be removed and open it. Then, * load its first sublist and determine which block pointers should actually * be freed. Then, call a synctask which performs the actual frees and updates * the pool-wide livelist data. */ static void spa_livelist_delete_cb(void *arg, zthr_t *z) { spa_t *spa = arg; uint64_t ll_obj = 0, count; objset_t *mos = spa->spa_meta_objset; uint64_t zap_obj = spa->spa_livelists_to_delete; /* * Determine the next livelist to delete. This function should only * be called if there is at least one deleted clone. */ VERIFY0(dsl_get_next_livelist_obj(mos, zap_obj, &ll_obj)); VERIFY0(zap_count(mos, ll_obj, &count)); if (count > 0) { dsl_deadlist_t *ll; dsl_deadlist_entry_t *dle; bplist_t to_free; ll = kmem_zalloc(sizeof (dsl_deadlist_t), KM_SLEEP); dsl_deadlist_open(ll, mos, ll_obj); dle = dsl_deadlist_first(ll); ASSERT3P(dle, !=, NULL); bplist_create(&to_free); int err = dsl_process_sub_livelist(&dle->dle_bpobj, &to_free, z, NULL); if (err == 0) { sublist_delete_arg_t sync_arg = { .spa = spa, .ll = ll, .key = dle->dle_mintxg, .to_free = &to_free }; zfs_dbgmsg("deleting sublist (id %llu) from" " livelist %llu, %lld remaining", (u_longlong_t)dle->dle_bpobj.bpo_object, (u_longlong_t)ll_obj, (longlong_t)count - 1); VERIFY0(dsl_sync_task(spa_name(spa), NULL, sublist_delete_sync, &sync_arg, 0, ZFS_SPACE_CHECK_DESTROY)); } else { VERIFY3U(err, ==, EINTR); } bplist_clear(&to_free); bplist_destroy(&to_free); dsl_deadlist_close(ll); kmem_free(ll, sizeof (dsl_deadlist_t)); } else { livelist_delete_arg_t sync_arg = { .spa = spa, .ll_obj = ll_obj, .zap_obj = zap_obj }; zfs_dbgmsg("deletion of livelist %llu completed", (u_longlong_t)ll_obj); VERIFY0(dsl_sync_task(spa_name(spa), NULL, livelist_delete_sync, &sync_arg, 0, ZFS_SPACE_CHECK_DESTROY)); } } static void spa_start_livelist_destroy_thread(spa_t *spa) { ASSERT3P(spa->spa_livelist_delete_zthr, ==, NULL); spa->spa_livelist_delete_zthr = zthr_create("z_livelist_destroy", spa_livelist_delete_cb_check, spa_livelist_delete_cb, spa, minclsyspri); } typedef struct livelist_new_arg { bplist_t *allocs; bplist_t *frees; } livelist_new_arg_t; static int livelist_track_new_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx) { ASSERT(tx == NULL); livelist_new_arg_t *lna = arg; if (bp_freed) { bplist_append(lna->frees, bp); } else { bplist_append(lna->allocs, bp); zfs_livelist_condense_new_alloc++; } return (0); } typedef struct livelist_condense_arg { spa_t *spa; bplist_t to_keep; uint64_t first_size; uint64_t next_size; } livelist_condense_arg_t; static void spa_livelist_condense_sync(void *arg, dmu_tx_t *tx) { livelist_condense_arg_t *lca = arg; spa_t *spa = lca->spa; bplist_t new_frees; dsl_dataset_t *ds = spa->spa_to_condense.ds; /* Have we been cancelled? */ if (spa->spa_to_condense.cancelled) { zfs_livelist_condense_sync_cancel++; goto out; } dsl_deadlist_entry_t *first = spa->spa_to_condense.first; dsl_deadlist_entry_t *next = spa->spa_to_condense.next; dsl_deadlist_t *ll = &ds->ds_dir->dd_livelist; /* * It's possible that the livelist was changed while the zthr was * running. Therefore, we need to check for new blkptrs in the two * entries being condensed and continue to track them in the livelist. * Because of the way we handle remapped blkptrs (see dbuf_remap_impl), * it's possible that the newly added blkptrs are FREEs or ALLOCs so * we need to sort them into two different bplists. */ uint64_t first_obj = first->dle_bpobj.bpo_object; uint64_t next_obj = next->dle_bpobj.bpo_object; uint64_t cur_first_size = first->dle_bpobj.bpo_phys->bpo_num_blkptrs; uint64_t cur_next_size = next->dle_bpobj.bpo_phys->bpo_num_blkptrs; bplist_create(&new_frees); livelist_new_arg_t new_bps = { .allocs = &lca->to_keep, .frees = &new_frees, }; if (cur_first_size > lca->first_size) { VERIFY0(livelist_bpobj_iterate_from_nofree(&first->dle_bpobj, livelist_track_new_cb, &new_bps, lca->first_size)); } if (cur_next_size > lca->next_size) { VERIFY0(livelist_bpobj_iterate_from_nofree(&next->dle_bpobj, livelist_track_new_cb, &new_bps, lca->next_size)); } dsl_deadlist_clear_entry(first, ll, tx); ASSERT(bpobj_is_empty(&first->dle_bpobj)); dsl_deadlist_remove_entry(ll, next->dle_mintxg, tx); bplist_iterate(&lca->to_keep, dsl_deadlist_insert_alloc_cb, ll, tx); bplist_iterate(&new_frees, dsl_deadlist_insert_free_cb, ll, tx); bplist_destroy(&new_frees); char dsname[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_name(ds, dsname); zfs_dbgmsg("txg %llu condensing livelist of %s (id %llu), bpobj %llu " "(%llu blkptrs) and bpobj %llu (%llu blkptrs) -> bpobj %llu " "(%llu blkptrs)", (u_longlong_t)tx->tx_txg, dsname, (u_longlong_t)ds->ds_object, (u_longlong_t)first_obj, (u_longlong_t)cur_first_size, (u_longlong_t)next_obj, (u_longlong_t)cur_next_size, (u_longlong_t)first->dle_bpobj.bpo_object, (u_longlong_t)first->dle_bpobj.bpo_phys->bpo_num_blkptrs); out: dmu_buf_rele(ds->ds_dbuf, spa); spa->spa_to_condense.ds = NULL; bplist_clear(&lca->to_keep); bplist_destroy(&lca->to_keep); kmem_free(lca, sizeof (livelist_condense_arg_t)); spa->spa_to_condense.syncing = B_FALSE; } static void spa_livelist_condense_cb(void *arg, zthr_t *t) { while (zfs_livelist_condense_zthr_pause && !(zthr_has_waiters(t) || zthr_iscancelled(t))) delay(1); spa_t *spa = arg; dsl_deadlist_entry_t *first = spa->spa_to_condense.first; dsl_deadlist_entry_t *next = spa->spa_to_condense.next; uint64_t first_size, next_size; livelist_condense_arg_t *lca = kmem_alloc(sizeof (livelist_condense_arg_t), KM_SLEEP); bplist_create(&lca->to_keep); /* * Process the livelists (matching FREEs and ALLOCs) in open context * so we have minimal work in syncing context to condense. * * We save bpobj sizes (first_size and next_size) to use later in * syncing context to determine if entries were added to these sublists * while in open context. This is possible because the clone is still * active and open for normal writes and we want to make sure the new, * unprocessed blockpointers are inserted into the livelist normally. * * Note that dsl_process_sub_livelist() both stores the size number of * blockpointers and iterates over them while the bpobj's lock held, so * the sizes returned to us are consistent which what was actually * processed. */ int err = dsl_process_sub_livelist(&first->dle_bpobj, &lca->to_keep, t, &first_size); if (err == 0) err = dsl_process_sub_livelist(&next->dle_bpobj, &lca->to_keep, t, &next_size); if (err == 0) { while (zfs_livelist_condense_sync_pause && !(zthr_has_waiters(t) || zthr_iscancelled(t))) delay(1); dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); dmu_tx_mark_netfree(tx); dmu_tx_hold_space(tx, 1); err = dmu_tx_assign(tx, TXG_NOWAIT | TXG_NOTHROTTLE); if (err == 0) { /* * Prevent the condense zthr restarting before * the synctask completes. */ spa->spa_to_condense.syncing = B_TRUE; lca->spa = spa; lca->first_size = first_size; lca->next_size = next_size; dsl_sync_task_nowait(spa_get_dsl(spa), spa_livelist_condense_sync, lca, tx); dmu_tx_commit(tx); return; } } /* * Condensing can not continue: either it was externally stopped or * we were unable to assign to a tx because the pool has run out of * space. In the second case, we'll just end up trying to condense * again in a later txg. */ ASSERT(err != 0); bplist_clear(&lca->to_keep); bplist_destroy(&lca->to_keep); kmem_free(lca, sizeof (livelist_condense_arg_t)); dmu_buf_rele(spa->spa_to_condense.ds->ds_dbuf, spa); spa->spa_to_condense.ds = NULL; if (err == EINTR) zfs_livelist_condense_zthr_cancel++; } /* * Check that there is something to condense but that a condense is not * already in progress and that condensing has not been cancelled. */ static boolean_t spa_livelist_condense_cb_check(void *arg, zthr_t *z) { (void) z; spa_t *spa = arg; if ((spa->spa_to_condense.ds != NULL) && (spa->spa_to_condense.syncing == B_FALSE) && (spa->spa_to_condense.cancelled == B_FALSE)) { return (B_TRUE); } return (B_FALSE); } static void spa_start_livelist_condensing_thread(spa_t *spa) { spa->spa_to_condense.ds = NULL; spa->spa_to_condense.first = NULL; spa->spa_to_condense.next = NULL; spa->spa_to_condense.syncing = B_FALSE; spa->spa_to_condense.cancelled = B_FALSE; ASSERT3P(spa->spa_livelist_condense_zthr, ==, NULL); spa->spa_livelist_condense_zthr = zthr_create("z_livelist_condense", spa_livelist_condense_cb_check, spa_livelist_condense_cb, spa, minclsyspri); } static void spa_spawn_aux_threads(spa_t *spa) { ASSERT(spa_writeable(spa)); ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_start_indirect_condensing_thread(spa); spa_start_livelist_destroy_thread(spa); spa_start_livelist_condensing_thread(spa); ASSERT3P(spa->spa_checkpoint_discard_zthr, ==, NULL); spa->spa_checkpoint_discard_zthr = zthr_create("z_checkpoint_discard", spa_checkpoint_discard_thread_check, spa_checkpoint_discard_thread, spa, minclsyspri); } /* * Fix up config after a partly-completed split. This is done with the * ZPOOL_CONFIG_SPLIT nvlist. Both the splitting pool and the split-off * pool have that entry in their config, but only the splitting one contains * a list of all the guids of the vdevs that are being split off. * * This function determines what to do with that list: either rejoin * all the disks to the pool, or complete the splitting process. To attempt * the rejoin, each disk that is offlined is marked online again, and * we do a reopen() call. If the vdev label for every disk that was * marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL) * then we call vdev_split() on each disk, and complete the split. * * Otherwise we leave the config alone, with all the vdevs in place in * the original pool. */ static void spa_try_repair(spa_t *spa, nvlist_t *config) { uint_t extracted; uint64_t *glist; uint_t i, gcount; nvlist_t *nvl; vdev_t **vd; boolean_t attempt_reopen; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0) return; /* check that the config is complete */ if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, &glist, &gcount) != 0) return; vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_SLEEP); /* attempt to online all the vdevs & validate */ attempt_reopen = B_TRUE; for (i = 0; i < gcount; i++) { if (glist[i] == 0) /* vdev is hole */ continue; vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE); if (vd[i] == NULL) { /* * Don't bother attempting to reopen the disks; * just do the split. */ attempt_reopen = B_FALSE; } else { /* attempt to re-online it */ vd[i]->vdev_offline = B_FALSE; } } if (attempt_reopen) { vdev_reopen(spa->spa_root_vdev); /* check each device to see what state it's in */ for (extracted = 0, i = 0; i < gcount; i++) { if (vd[i] != NULL && vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL) break; ++extracted; } } /* * If every disk has been moved to the new pool, or if we never * even attempted to look at them, then we split them off for * good. */ if (!attempt_reopen || gcount == extracted) { for (i = 0; i < gcount; i++) if (vd[i] != NULL) vdev_split(vd[i]); vdev_reopen(spa->spa_root_vdev); } kmem_free(vd, gcount * sizeof (vdev_t *)); } static int spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type) { const char *ereport = FM_EREPORT_ZFS_POOL; int error; spa->spa_load_state = state; (void) spa_import_progress_set_state(spa_guid(spa), spa_load_state(spa)); gethrestime(&spa->spa_loaded_ts); error = spa_load_impl(spa, type, &ereport); /* * Don't count references from objsets that are already closed * and are making their way through the eviction process. */ spa_evicting_os_wait(spa); spa->spa_minref = zfs_refcount_count(&spa->spa_refcount); if (error) { if (error != EEXIST) { spa->spa_loaded_ts.tv_sec = 0; spa->spa_loaded_ts.tv_nsec = 0; } if (error != EBADF) { (void) zfs_ereport_post(ereport, spa, NULL, NULL, NULL, 0); } } spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE; spa->spa_ena = 0; (void) spa_import_progress_set_state(spa_guid(spa), spa_load_state(spa)); return (error); } #ifdef ZFS_DEBUG /* * Count the number of per-vdev ZAPs associated with all of the vdevs in the * vdev tree rooted in the given vd, and ensure that each ZAP is present in the * spa's per-vdev ZAP list. */ static uint64_t vdev_count_verify_zaps(vdev_t *vd) { spa_t *spa = vd->vdev_spa; uint64_t total = 0; if (vd->vdev_top_zap != 0) { total++; ASSERT0(zap_lookup_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, vd->vdev_top_zap)); } if (vd->vdev_leaf_zap != 0) { total++; ASSERT0(zap_lookup_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, vd->vdev_leaf_zap)); } for (uint64_t i = 0; i < vd->vdev_children; i++) { total += vdev_count_verify_zaps(vd->vdev_child[i]); } return (total); } #else #define vdev_count_verify_zaps(vd) ((void) sizeof (vd), 0) #endif /* * Determine whether the activity check is required. */ static boolean_t spa_activity_check_required(spa_t *spa, uberblock_t *ub, nvlist_t *label, nvlist_t *config) { uint64_t state = 0; uint64_t hostid = 0; uint64_t tryconfig_txg = 0; uint64_t tryconfig_timestamp = 0; uint16_t tryconfig_mmp_seq = 0; nvlist_t *nvinfo; if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) { nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); (void) nvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG, &tryconfig_txg); (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_TIMESTAMP, &tryconfig_timestamp); (void) nvlist_lookup_uint16(nvinfo, ZPOOL_CONFIG_MMP_SEQ, &tryconfig_mmp_seq); } (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, &state); /* * Disable the MMP activity check - This is used by zdb which * is intended to be used on potentially active pools. */ if (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP) return (B_FALSE); /* * Skip the activity check when the MMP feature is disabled. */ if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay == 0) return (B_FALSE); /* * If the tryconfig_ values are nonzero, they are the results of an * earlier tryimport. If they all match the uberblock we just found, * then the pool has not changed and we return false so we do not test * a second time. */ if (tryconfig_txg && tryconfig_txg == ub->ub_txg && tryconfig_timestamp && tryconfig_timestamp == ub->ub_timestamp && tryconfig_mmp_seq && tryconfig_mmp_seq == (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0)) return (B_FALSE); /* * Allow the activity check to be skipped when importing the pool * on the same host which last imported it. Since the hostid from * configuration may be stale use the one read from the label. */ if (nvlist_exists(label, ZPOOL_CONFIG_HOSTID)) hostid = fnvlist_lookup_uint64(label, ZPOOL_CONFIG_HOSTID); if (hostid == spa_get_hostid(spa)) return (B_FALSE); /* * Skip the activity test when the pool was cleanly exported. */ if (state != POOL_STATE_ACTIVE) return (B_FALSE); return (B_TRUE); } /* * Nanoseconds the activity check must watch for changes on-disk. */ static uint64_t spa_activity_check_duration(spa_t *spa, uberblock_t *ub) { uint64_t import_intervals = MAX(zfs_multihost_import_intervals, 1); uint64_t multihost_interval = MSEC2NSEC( MMP_INTERVAL_OK(zfs_multihost_interval)); uint64_t import_delay = MAX(NANOSEC, import_intervals * multihost_interval); /* * Local tunables determine a minimum duration except for the case * where we know when the remote host will suspend the pool if MMP * writes do not land. * * See Big Theory comment at the top of mmp.c for the reasoning behind * these cases and times. */ ASSERT(MMP_IMPORT_SAFETY_FACTOR >= 100); if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) && MMP_FAIL_INT(ub) > 0) { /* MMP on remote host will suspend pool after failed writes */ import_delay = MMP_FAIL_INT(ub) * MSEC2NSEC(MMP_INTERVAL(ub)) * MMP_IMPORT_SAFETY_FACTOR / 100; zfs_dbgmsg("fail_intvals>0 import_delay=%llu ub_mmp " "mmp_fails=%llu ub_mmp mmp_interval=%llu " "import_intervals=%llu", (u_longlong_t)import_delay, (u_longlong_t)MMP_FAIL_INT(ub), (u_longlong_t)MMP_INTERVAL(ub), (u_longlong_t)import_intervals); } else if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) && MMP_FAIL_INT(ub) == 0) { /* MMP on remote host will never suspend pool */ import_delay = MAX(import_delay, (MSEC2NSEC(MMP_INTERVAL(ub)) + ub->ub_mmp_delay) * import_intervals); zfs_dbgmsg("fail_intvals=0 import_delay=%llu ub_mmp " "mmp_interval=%llu ub_mmp_delay=%llu " "import_intervals=%llu", (u_longlong_t)import_delay, (u_longlong_t)MMP_INTERVAL(ub), (u_longlong_t)ub->ub_mmp_delay, (u_longlong_t)import_intervals); } else if (MMP_VALID(ub)) { /* * zfs-0.7 compatibility case */ import_delay = MAX(import_delay, (multihost_interval + ub->ub_mmp_delay) * import_intervals); zfs_dbgmsg("import_delay=%llu ub_mmp_delay=%llu " "import_intervals=%llu leaves=%u", (u_longlong_t)import_delay, (u_longlong_t)ub->ub_mmp_delay, (u_longlong_t)import_intervals, vdev_count_leaves(spa)); } else { /* Using local tunings is the only reasonable option */ zfs_dbgmsg("pool last imported on non-MMP aware " "host using import_delay=%llu multihost_interval=%llu " "import_intervals=%llu", (u_longlong_t)import_delay, (u_longlong_t)multihost_interval, (u_longlong_t)import_intervals); } return (import_delay); } /* * Perform the import activity check. If the user canceled the import or * we detected activity then fail. */ static int spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config) { uint64_t txg = ub->ub_txg; uint64_t timestamp = ub->ub_timestamp; uint64_t mmp_config = ub->ub_mmp_config; uint16_t mmp_seq = MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0; uint64_t import_delay; hrtime_t import_expire; nvlist_t *mmp_label = NULL; vdev_t *rvd = spa->spa_root_vdev; kcondvar_t cv; kmutex_t mtx; int error = 0; cv_init(&cv, NULL, CV_DEFAULT, NULL); mutex_init(&mtx, NULL, MUTEX_DEFAULT, NULL); mutex_enter(&mtx); /* * If ZPOOL_CONFIG_MMP_TXG is present an activity check was performed * during the earlier tryimport. If the txg recorded there is 0 then * the pool is known to be active on another host. * * Otherwise, the pool might be in use on another host. Check for * changes in the uberblocks on disk if necessary. */ if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) { nvlist_t *nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_TXG) && fnvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG) == 0) { vdev_uberblock_load(rvd, ub, &mmp_label); error = SET_ERROR(EREMOTEIO); goto out; } } import_delay = spa_activity_check_duration(spa, ub); /* Add a small random factor in case of simultaneous imports (0-25%) */ import_delay += import_delay * random_in_range(250) / 1000; import_expire = gethrtime() + import_delay; while (gethrtime() < import_expire) { (void) spa_import_progress_set_mmp_check(spa_guid(spa), NSEC2SEC(import_expire - gethrtime())); vdev_uberblock_load(rvd, ub, &mmp_label); if (txg != ub->ub_txg || timestamp != ub->ub_timestamp || mmp_seq != (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0)) { zfs_dbgmsg("multihost activity detected " "txg %llu ub_txg %llu " "timestamp %llu ub_timestamp %llu " "mmp_config %#llx ub_mmp_config %#llx", (u_longlong_t)txg, (u_longlong_t)ub->ub_txg, (u_longlong_t)timestamp, (u_longlong_t)ub->ub_timestamp, (u_longlong_t)mmp_config, (u_longlong_t)ub->ub_mmp_config); error = SET_ERROR(EREMOTEIO); break; } if (mmp_label) { nvlist_free(mmp_label); mmp_label = NULL; } error = cv_timedwait_sig(&cv, &mtx, ddi_get_lbolt() + hz); if (error != -1) { error = SET_ERROR(EINTR); break; } error = 0; } out: mutex_exit(&mtx); mutex_destroy(&mtx); cv_destroy(&cv); /* * If the pool is determined to be active store the status in the * spa->spa_load_info nvlist. If the remote hostname or hostid are * available from configuration read from disk store them as well. * This allows 'zpool import' to generate a more useful message. * * ZPOOL_CONFIG_MMP_STATE - observed pool status (mandatory) * ZPOOL_CONFIG_MMP_HOSTNAME - hostname from the active pool * ZPOOL_CONFIG_MMP_HOSTID - hostid from the active pool */ if (error == EREMOTEIO) { const char *hostname = ""; uint64_t hostid = 0; if (mmp_label) { if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTNAME)) { hostname = fnvlist_lookup_string(mmp_label, ZPOOL_CONFIG_HOSTNAME); fnvlist_add_string(spa->spa_load_info, ZPOOL_CONFIG_MMP_HOSTNAME, hostname); } if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTID)) { hostid = fnvlist_lookup_uint64(mmp_label, ZPOOL_CONFIG_HOSTID); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_HOSTID, hostid); } } fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_STATE, MMP_STATE_ACTIVE); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_TXG, 0); error = spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO); } if (mmp_label) nvlist_free(mmp_label); return (error); } static int spa_verify_host(spa_t *spa, nvlist_t *mos_config) { uint64_t hostid; char *hostname; uint64_t myhostid = 0; if (!spa_is_root(spa) && nvlist_lookup_uint64(mos_config, ZPOOL_CONFIG_HOSTID, &hostid) == 0) { hostname = fnvlist_lookup_string(mos_config, ZPOOL_CONFIG_HOSTNAME); myhostid = zone_get_hostid(NULL); if (hostid != 0 && myhostid != 0 && hostid != myhostid) { cmn_err(CE_WARN, "pool '%s' could not be " "loaded as it was last accessed by " "another system (host: %s hostid: 0x%llx). " "See: https://openzfs.github.io/openzfs-docs/msg/" "ZFS-8000-EY", spa_name(spa), hostname, (u_longlong_t)hostid); spa_load_failed(spa, "hostid verification failed: pool " "last accessed by host: %s (hostid: 0x%llx)", hostname, (u_longlong_t)hostid); return (SET_ERROR(EBADF)); } } return (0); } static int spa_ld_parse_config(spa_t *spa, spa_import_type_t type) { int error = 0; nvlist_t *nvtree, *nvl, *config = spa->spa_config; int parse; vdev_t *rvd; uint64_t pool_guid; char *comment; char *compatibility; /* * Versioning wasn't explicitly added to the label until later, so if * it's not present treat it as the initial version. */ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &spa->spa_ubsync.ub_version) != 0) spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL; if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) { spa_load_failed(spa, "invalid config provided: '%s' missing", ZPOOL_CONFIG_POOL_GUID); return (SET_ERROR(EINVAL)); } /* * If we are doing an import, ensure that the pool is not already * imported by checking if its pool guid already exists in the * spa namespace. * * The only case that we allow an already imported pool to be * imported again, is when the pool is checkpointed and we want to * look at its checkpointed state from userland tools like zdb. */ #ifdef _KERNEL if ((spa->spa_load_state == SPA_LOAD_IMPORT || spa->spa_load_state == SPA_LOAD_TRYIMPORT) && spa_guid_exists(pool_guid, 0)) { #else if ((spa->spa_load_state == SPA_LOAD_IMPORT || spa->spa_load_state == SPA_LOAD_TRYIMPORT) && spa_guid_exists(pool_guid, 0) && !spa_importing_readonly_checkpoint(spa)) { #endif spa_load_failed(spa, "a pool with guid %llu is already open", (u_longlong_t)pool_guid); return (SET_ERROR(EEXIST)); } spa->spa_config_guid = pool_guid; nvlist_free(spa->spa_load_info); spa->spa_load_info = fnvlist_alloc(); ASSERT(spa->spa_comment == NULL); if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0) spa->spa_comment = spa_strdup(comment); ASSERT(spa->spa_compatibility == NULL); if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMPATIBILITY, &compatibility) == 0) spa->spa_compatibility = spa_strdup(compatibility); (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &spa->spa_config_txg); if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0) spa->spa_config_splitting = fnvlist_dup(nvl); if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtree)) { spa_load_failed(spa, "invalid config provided: '%s' missing", ZPOOL_CONFIG_VDEV_TREE); return (SET_ERROR(EINVAL)); } /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *), KM_SLEEP); for (int i = 0; i < max_ncpus; i++) { spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); } /* * Parse the configuration into a vdev tree. We explicitly set the * value that will be returned by spa_version() since parsing the * configuration requires knowing the version number. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); parse = (type == SPA_IMPORT_EXISTING ? VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT); error = spa_config_parse(spa, &rvd, nvtree, NULL, 0, parse); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_load_failed(spa, "unable to parse config [error=%d]", error); return (error); } ASSERT(spa->spa_root_vdev == rvd); ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT); ASSERT3U(spa->spa_max_ashift, <=, SPA_MAXBLOCKSHIFT); if (type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_guid(spa) == pool_guid); } return (0); } /* * Recursively open all vdevs in the vdev tree. This function is called twice: * first with the untrusted config, then with the trusted config. */ static int spa_ld_open_vdevs(spa_t *spa) { int error = 0; /* * spa_missing_tvds_allowed defines how many top-level vdevs can be * missing/unopenable for the root vdev to be still considered openable. */ if (spa->spa_trust_config) { spa->spa_missing_tvds_allowed = zfs_max_missing_tvds; } else if (spa->spa_config_source == SPA_CONFIG_SRC_CACHEFILE) { spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_cachefile; } else if (spa->spa_config_source == SPA_CONFIG_SRC_SCAN) { spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_scan; } else { spa->spa_missing_tvds_allowed = 0; } spa->spa_missing_tvds_allowed = MAX(zfs_max_missing_tvds, spa->spa_missing_tvds_allowed); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_open(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); if (spa->spa_missing_tvds != 0) { spa_load_note(spa, "vdev tree has %lld missing top-level " "vdevs.", (u_longlong_t)spa->spa_missing_tvds); if (spa->spa_trust_config && (spa->spa_mode & SPA_MODE_WRITE)) { /* * Although theoretically we could allow users to open * incomplete pools in RW mode, we'd need to add a lot * of extra logic (e.g. adjust pool space to account * for missing vdevs). * This limitation also prevents users from accidentally * opening the pool in RW mode during data recovery and * damaging it further. */ spa_load_note(spa, "pools with missing top-level " "vdevs can only be opened in read-only mode."); error = SET_ERROR(ENXIO); } else { spa_load_note(spa, "current settings allow for maximum " "%lld missing top-level vdevs at this stage.", (u_longlong_t)spa->spa_missing_tvds_allowed); } } if (error != 0) { spa_load_failed(spa, "unable to open vdev tree [error=%d]", error); } if (spa->spa_missing_tvds != 0 || error != 0) vdev_dbgmsg_print_tree(spa->spa_root_vdev, 2); return (error); } /* * We need to validate the vdev labels against the configuration that * we have in hand. This function is called twice: first with an untrusted * config, then with a trusted config. The validation is more strict when the * config is trusted. */ static int spa_ld_validate_vdevs(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_validate(rvd); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_load_failed(spa, "vdev_validate failed [error=%d]", error); return (error); } if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) { spa_load_failed(spa, "cannot open vdev tree after invalidating " "some vdevs"); vdev_dbgmsg_print_tree(rvd, 2); return (SET_ERROR(ENXIO)); } return (0); } static void spa_ld_select_uberblock_done(spa_t *spa, uberblock_t *ub) { spa->spa_state = POOL_STATE_ACTIVE; spa->spa_ubsync = spa->spa_uberblock; spa->spa_verify_min_txg = spa->spa_extreme_rewind ? TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1; spa->spa_first_txg = spa->spa_last_ubsync_txg ? spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1; spa->spa_claim_max_txg = spa->spa_first_txg; spa->spa_prev_software_version = ub->ub_software_version; } static int spa_ld_select_uberblock(spa_t *spa, spa_import_type_t type) { vdev_t *rvd = spa->spa_root_vdev; nvlist_t *label; uberblock_t *ub = &spa->spa_uberblock; boolean_t activity_check = B_FALSE; /* * If we are opening the checkpointed state of the pool by * rewinding to it, at this point we will have written the * checkpointed uberblock to the vdev labels, so searching * the labels will find the right uberblock. However, if * we are opening the checkpointed state read-only, we have * not modified the labels. Therefore, we must ignore the * labels and continue using the spa_uberblock that was set * by spa_ld_checkpoint_rewind. * * Note that it would be fine to ignore the labels when * rewinding (opening writeable) as well. However, if we * crash just after writing the labels, we will end up * searching the labels. Doing so in the common case means * that this code path gets exercised normally, rather than * just in the edge case. */ if (ub->ub_checkpoint_txg != 0 && spa_importing_readonly_checkpoint(spa)) { spa_ld_select_uberblock_done(spa, ub); return (0); } /* * Find the best uberblock. */ vdev_uberblock_load(rvd, ub, &label); /* * If we weren't able to find a single valid uberblock, return failure. */ if (ub->ub_txg == 0) { nvlist_free(label); spa_load_failed(spa, "no valid uberblock found"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } if (spa->spa_load_max_txg != UINT64_MAX) { (void) spa_import_progress_set_max_txg(spa_guid(spa), (u_longlong_t)spa->spa_load_max_txg); } spa_load_note(spa, "using uberblock with txg=%llu", (u_longlong_t)ub->ub_txg); /* * For pools which have the multihost property on determine if the * pool is truly inactive and can be safely imported. Prevent * hosts which don't have a hostid set from importing the pool. */ activity_check = spa_activity_check_required(spa, ub, label, spa->spa_config); if (activity_check) { if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay && spa_get_hostid(spa) == 0) { nvlist_free(label); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID); return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO)); } int error = spa_activity_check(spa, ub, spa->spa_config); if (error) { nvlist_free(label); return (error); } fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_STATE, MMP_STATE_INACTIVE); fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_TXG, ub->ub_txg); fnvlist_add_uint16(spa->spa_load_info, ZPOOL_CONFIG_MMP_SEQ, (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0)); } /* * If the pool has an unsupported version we can't open it. */ if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) { nvlist_free(label); spa_load_failed(spa, "version %llu is not supported", (u_longlong_t)ub->ub_version); return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP)); } if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *features; /* * If we weren't able to find what's necessary for reading the * MOS in the label, return failure. */ if (label == NULL) { spa_load_failed(spa, "label config unavailable"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) != 0) { nvlist_free(label); spa_load_failed(spa, "invalid label: '%s' missing", ZPOOL_CONFIG_FEATURES_FOR_READ); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } /* * Update our in-core representation with the definitive values * from the label. */ nvlist_free(spa->spa_label_features); spa->spa_label_features = fnvlist_dup(features); } nvlist_free(label); /* * Look through entries in the label nvlist's features_for_read. If * there is a feature listed there which we don't understand then we * cannot open a pool. */ if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *unsup_feat; unsup_feat = fnvlist_alloc(); for (nvpair_t *nvp = nvlist_next_nvpair(spa->spa_label_features, NULL); nvp != NULL; nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) { if (!zfeature_is_supported(nvpair_name(nvp))) { fnvlist_add_string(unsup_feat, nvpair_name(nvp), ""); } } if (!nvlist_empty(unsup_feat)) { fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat); nvlist_free(unsup_feat); spa_load_failed(spa, "some features are unsupported"); return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } nvlist_free(unsup_feat); } if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_try_repair(spa, spa->spa_config); spa_config_exit(spa, SCL_ALL, FTAG); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; } /* * Initialize internal SPA structures. */ spa_ld_select_uberblock_done(spa, ub); return (0); } static int spa_ld_open_rootbp(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool); if (error != 0) { spa_load_failed(spa, "unable to open rootbp in dsl_pool_init " "[error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset; return (0); } static int spa_ld_trusted_config(spa_t *spa, spa_import_type_t type, boolean_t reloading) { vdev_t *mrvd, *rvd = spa->spa_root_vdev; nvlist_t *nv, *mos_config, *policy; int error = 0, copy_error; uint64_t healthy_tvds, healthy_tvds_mos; uint64_t mos_config_txg; if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object, B_TRUE) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * If we're assembling a pool from a split, the config provided is * already trusted so there is nothing to do. */ if (type == SPA_IMPORT_ASSEMBLE) return (0); healthy_tvds = spa_healthy_core_tvds(spa); if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0) { spa_load_failed(spa, "unable to retrieve MOS config"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } /* * If we are doing an open, pool owner wasn't verified yet, thus do * the verification here. */ if (spa->spa_load_state == SPA_LOAD_OPEN) { error = spa_verify_host(spa, mos_config); if (error != 0) { nvlist_free(mos_config); return (error); } } nv = fnvlist_lookup_nvlist(mos_config, ZPOOL_CONFIG_VDEV_TREE); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Build a new vdev tree from the trusted config */ error = spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD); if (error != 0) { nvlist_free(mos_config); spa_config_exit(spa, SCL_ALL, FTAG); spa_load_failed(spa, "spa_config_parse failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } /* * Vdev paths in the MOS may be obsolete. If the untrusted config was * obtained by scanning /dev/dsk, then it will have the right vdev * paths. We update the trusted MOS config with this information. * We first try to copy the paths with vdev_copy_path_strict, which * succeeds only when both configs have exactly the same vdev tree. * If that fails, we fall back to a more flexible method that has a * best effort policy. */ copy_error = vdev_copy_path_strict(rvd, mrvd); if (copy_error != 0 || spa_load_print_vdev_tree) { spa_load_note(spa, "provided vdev tree:"); vdev_dbgmsg_print_tree(rvd, 2); spa_load_note(spa, "MOS vdev tree:"); vdev_dbgmsg_print_tree(mrvd, 2); } if (copy_error != 0) { spa_load_note(spa, "vdev_copy_path_strict failed, falling " "back to vdev_copy_path_relaxed"); vdev_copy_path_relaxed(rvd, mrvd); } vdev_close(rvd); vdev_free(rvd); spa->spa_root_vdev = mrvd; rvd = mrvd; spa_config_exit(spa, SCL_ALL, FTAG); /* * We will use spa_config if we decide to reload the spa or if spa_load * fails and we rewind. We must thus regenerate the config using the * MOS information with the updated paths. ZPOOL_LOAD_POLICY is used to * pass settings on how to load the pool and is not stored in the MOS. * We copy it over to our new, trusted config. */ mos_config_txg = fnvlist_lookup_uint64(mos_config, ZPOOL_CONFIG_POOL_TXG); nvlist_free(mos_config); mos_config = spa_config_generate(spa, NULL, mos_config_txg, B_FALSE); if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_LOAD_POLICY, &policy) == 0) fnvlist_add_nvlist(mos_config, ZPOOL_LOAD_POLICY, policy); spa_config_set(spa, mos_config); spa->spa_config_source = SPA_CONFIG_SRC_MOS; /* * Now that we got the config from the MOS, we should be more strict * in checking blkptrs and can make assumptions about the consistency * of the vdev tree. spa_trust_config must be set to true before opening * vdevs in order for them to be writeable. */ spa->spa_trust_config = B_TRUE; /* * Open and validate the new vdev tree */ error = spa_ld_open_vdevs(spa); if (error != 0) return (error); error = spa_ld_validate_vdevs(spa); if (error != 0) return (error); if (copy_error != 0 || spa_load_print_vdev_tree) { spa_load_note(spa, "final vdev tree:"); vdev_dbgmsg_print_tree(rvd, 2); } if (spa->spa_load_state != SPA_LOAD_TRYIMPORT && !spa->spa_extreme_rewind && zfs_max_missing_tvds == 0) { /* * Sanity check to make sure that we are indeed loading the * latest uberblock. If we missed SPA_SYNC_MIN_VDEVS tvds * in the config provided and they happened to be the only ones * to have the latest uberblock, we could involuntarily perform * an extreme rewind. */ healthy_tvds_mos = spa_healthy_core_tvds(spa); if (healthy_tvds_mos - healthy_tvds >= SPA_SYNC_MIN_VDEVS) { spa_load_note(spa, "config provided misses too many " "top-level vdevs compared to MOS (%lld vs %lld). ", (u_longlong_t)healthy_tvds, (u_longlong_t)healthy_tvds_mos); spa_load_note(spa, "vdev tree:"); vdev_dbgmsg_print_tree(rvd, 2); if (reloading) { spa_load_failed(spa, "config was already " "provided from MOS. Aborting."); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa_load_note(spa, "spa must be reloaded using MOS " "config"); return (SET_ERROR(EAGAIN)); } } error = spa_check_for_missing_logs(spa); if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); if (rvd->vdev_guid_sum != spa->spa_uberblock.ub_guid_sum) { spa_load_failed(spa, "uberblock guid sum doesn't match MOS " "guid sum (%llu != %llu)", (u_longlong_t)spa->spa_uberblock.ub_guid_sum, (u_longlong_t)rvd->vdev_guid_sum); return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); } return (0); } static int spa_ld_open_indirect_vdev_metadata(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; /* * Everything that we read before spa_remove_init() must be stored * on concreted vdevs. Therefore we do this as early as possible. */ error = spa_remove_init(spa); if (error != 0) { spa_load_failed(spa, "spa_remove_init failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } /* * Retrieve information needed to condense indirect vdev mappings. */ error = spa_condense_init(spa); if (error != 0) { spa_load_failed(spa, "spa_condense_init failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } return (0); } static int spa_ld_check_features(spa_t *spa, boolean_t *missing_feat_writep) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; if (spa_version(spa) >= SPA_VERSION_FEATURES) { boolean_t missing_feat_read = B_FALSE; nvlist_t *unsup_feat, *enabled_feat; if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ, &spa->spa_feat_for_read_obj, B_TRUE) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_WRITE, &spa->spa_feat_for_write_obj, B_TRUE) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURE_DESCRIPTIONS, &spa->spa_feat_desc_obj, B_TRUE) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } enabled_feat = fnvlist_alloc(); unsup_feat = fnvlist_alloc(); if (!spa_features_check(spa, B_FALSE, unsup_feat, enabled_feat)) missing_feat_read = B_TRUE; if (spa_writeable(spa) || spa->spa_load_state == SPA_LOAD_TRYIMPORT) { if (!spa_features_check(spa, B_TRUE, unsup_feat, enabled_feat)) { *missing_feat_writep = B_TRUE; } } fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_ENABLED_FEAT, enabled_feat); if (!nvlist_empty(unsup_feat)) { fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat); } fnvlist_free(enabled_feat); fnvlist_free(unsup_feat); if (!missing_feat_read) { fnvlist_add_boolean(spa->spa_load_info, ZPOOL_CONFIG_CAN_RDONLY); } /* * If the state is SPA_LOAD_TRYIMPORT, our objective is * twofold: to determine whether the pool is available for * import in read-write mode and (if it is not) whether the * pool is available for import in read-only mode. If the pool * is available for import in read-write mode, it is displayed * as available in userland; if it is not available for import * in read-only mode, it is displayed as unavailable in * userland. If the pool is available for import in read-only * mode but not read-write mode, it is displayed as unavailable * in userland with a special note that the pool is actually * available for open in read-only mode. * * As a result, if the state is SPA_LOAD_TRYIMPORT and we are * missing a feature for write, we must first determine whether * the pool can be opened read-only before returning to * userland in order to know whether to display the * abovementioned note. */ if (missing_feat_read || (*missing_feat_writep && spa_writeable(spa))) { spa_load_failed(spa, "pool uses unsupported features"); return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * Load refcounts for ZFS features from disk into an in-memory * cache during SPA initialization. */ for (spa_feature_t i = 0; i < SPA_FEATURES; i++) { uint64_t refcount; error = feature_get_refcount_from_disk(spa, &spa_feature_table[i], &refcount); if (error == 0) { spa->spa_feat_refcount_cache[i] = refcount; } else if (error == ENOTSUP) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } else { spa_load_failed(spa, "error getting refcount " "for feature %s [error=%d]", spa_feature_table[i].fi_guid, error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } } } if (spa_feature_is_active(spa, SPA_FEATURE_ENABLED_TXG)) { if (spa_dir_prop(spa, DMU_POOL_FEATURE_ENABLED_TXG, &spa->spa_feat_enabled_txg_obj, B_TRUE) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } /* * Encryption was added before bookmark_v2, even though bookmark_v2 * is now a dependency. If this pool has encryption enabled without * bookmark_v2, trigger an errata message. */ if (spa_feature_is_enabled(spa, SPA_FEATURE_ENCRYPTION) && !spa_feature_is_enabled(spa, SPA_FEATURE_BOOKMARK_V2)) { spa->spa_errata = ZPOOL_ERRATA_ZOL_8308_ENCRYPTION; } return (0); } static int spa_ld_load_special_directories(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; spa->spa_is_initializing = B_TRUE; error = dsl_pool_open(spa->spa_dsl_pool); spa->spa_is_initializing = B_FALSE; if (error != 0) { spa_load_failed(spa, "dsl_pool_open failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } return (0); } static int spa_ld_get_props(spa_t *spa) { int error = 0; uint64_t obj; vdev_t *rvd = spa->spa_root_vdev; /* Grab the checksum salt from the MOS. */ error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CHECKSUM_SALT, 1, sizeof (spa->spa_cksum_salt.zcs_bytes), spa->spa_cksum_salt.zcs_bytes); if (error == ENOENT) { /* Generate a new salt for subsequent use */ (void) random_get_pseudo_bytes(spa->spa_cksum_salt.zcs_bytes, sizeof (spa->spa_cksum_salt.zcs_bytes)); } else if (error != 0) { spa_load_failed(spa, "unable to retrieve checksum salt from " "MOS [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj, B_TRUE) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj); if (error != 0) { spa_load_failed(spa, "error opening deferred-frees bpobj " "[error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } /* * Load the bit that tells us to use the new accounting function * (raid-z deflation). If we have an older pool, this will not * be present. */ error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_CREATION_VERSION, &spa->spa_creation_version, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the persistent error log. If we have an older pool, this will * not be present. */ error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_ERRLOG_SCRUB, &spa->spa_errlog_scrub, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the livelist deletion field. If a livelist is queued for * deletion, indicate that in the spa */ error = spa_dir_prop(spa, DMU_POOL_DELETED_CLONES, &spa->spa_livelists_to_delete, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the history object. If we have an older pool, this * will not be present. */ error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the per-vdev ZAP map. If we have an older pool, this will not * be present; in this case, defer its creation to a later time to * avoid dirtying the MOS this early / out of sync context. See * spa_sync_config_object. */ /* The sentinel is only available in the MOS config. */ nvlist_t *mos_config; if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0) { spa_load_failed(spa, "unable to retrieve MOS config"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } error = spa_dir_prop(spa, DMU_POOL_VDEV_ZAP_MAP, &spa->spa_all_vdev_zaps, B_FALSE); if (error == ENOENT) { VERIFY(!nvlist_exists(mos_config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS)); spa->spa_avz_action = AVZ_ACTION_INITIALIZE; ASSERT0(vdev_count_verify_zaps(spa->spa_root_vdev)); } else if (error != 0) { nvlist_free(mos_config); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } else if (!nvlist_exists(mos_config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS)) { /* * An older version of ZFS overwrote the sentinel value, so * we have orphaned per-vdev ZAPs in the MOS. Defer their * destruction to later; see spa_sync_config_object. */ spa->spa_avz_action = AVZ_ACTION_DESTROY; /* * We're assuming that no vdevs have had their ZAPs created * before this. Better be sure of it. */ ASSERT0(vdev_count_verify_zaps(spa->spa_root_vdev)); } nvlist_free(mos_config); spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object, B_FALSE); if (error && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0) { uint64_t autoreplace = 0; spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs); spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace); spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation); spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode); spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand); spa_prop_find(spa, ZPOOL_PROP_MULTIHOST, &spa->spa_multihost); spa_prop_find(spa, ZPOOL_PROP_AUTOTRIM, &spa->spa_autotrim); spa->spa_autoreplace = (autoreplace != 0); } /* * If we are importing a pool with missing top-level vdevs, * we enforce that the pool doesn't panic or get suspended on * error since the likelihood of missing data is extremely high. */ if (spa->spa_missing_tvds > 0 && spa->spa_failmode != ZIO_FAILURE_MODE_CONTINUE && spa->spa_load_state != SPA_LOAD_TRYIMPORT) { spa_load_note(spa, "forcing failmode to 'continue' " "as some top level vdevs are missing"); spa->spa_failmode = ZIO_FAILURE_MODE_CONTINUE; } return (0); } static int spa_ld_open_aux_vdevs(spa_t *spa, spa_import_type_t type) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; /* * If we're assembling the pool from the split-off vdevs of * an existing pool, we don't want to attach the spares & cache * devices. */ /* * Load any hot spares for this pool. */ error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_SPARES); if (load_nvlist(spa, spa->spa_spares.sav_object, &spa->spa_spares.sav_config) != 0) { spa_load_failed(spa, "error loading spares nvlist"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_spares.sav_sync = B_TRUE; } /* * Load any level 2 ARC devices for this pool. */ error = spa_dir_prop(spa, DMU_POOL_L2CACHE, &spa->spa_l2cache.sav_object, B_FALSE); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE); if (load_nvlist(spa, spa->spa_l2cache.sav_object, &spa->spa_l2cache.sav_config) != 0) { spa_load_failed(spa, "error loading l2cache nvlist"); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_l2cache.sav_sync = B_TRUE; } return (0); } static int spa_ld_load_vdev_metadata(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; /* * If the 'multihost' property is set, then never allow a pool to * be imported when the system hostid is zero. The exception to * this rule is zdb which is always allowed to access pools. */ if (spa_multihost(spa) && spa_get_hostid(spa) == 0 && (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP) == 0) { fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID); return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO)); } /* * If the 'autoreplace' property is set, then post a resource notifying * the ZFS DE that it should not issue any faults for unopenable * devices. We also iterate over the vdevs, and post a sysevent for any * unopenable vdevs so that the normal autoreplace handler can take * over. */ if (spa->spa_autoreplace && spa->spa_load_state != SPA_LOAD_TRYIMPORT) { spa_check_removed(spa->spa_root_vdev); /* * For the import case, this is done in spa_import(), because * at this point we're using the spare definitions from * the MOS config, not necessarily from the userland config. */ if (spa->spa_load_state != SPA_LOAD_IMPORT) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } } /* * Load the vdev metadata such as metaslabs, DTLs, spacemap object, etc. */ error = vdev_load(rvd); if (error != 0) { spa_load_failed(spa, "vdev_load failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } error = spa_ld_log_spacemaps(spa); if (error != 0) { spa_load_failed(spa, "spa_ld_log_spacemaps failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } /* * Propagate the leaf DTLs we just loaded all the way up the vdev tree. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_dtl_reassess(rvd, 0, 0, B_FALSE, B_FALSE); spa_config_exit(spa, SCL_ALL, FTAG); return (0); } static int spa_ld_load_dedup_tables(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; error = ddt_load(spa); if (error != 0) { spa_load_failed(spa, "ddt_load failed [error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } return (0); } static int spa_ld_verify_logs(spa_t *spa, spa_import_type_t type, const char **ereport) { vdev_t *rvd = spa->spa_root_vdev; if (type != SPA_IMPORT_ASSEMBLE && spa_writeable(spa)) { boolean_t missing = spa_check_logs(spa); if (missing) { if (spa->spa_missing_tvds != 0) { spa_load_note(spa, "spa_check_logs failed " "so dropping the logs"); } else { *ereport = FM_EREPORT_ZFS_LOG_REPLAY; spa_load_failed(spa, "spa_check_logs failed"); return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG, ENXIO)); } } } return (0); } static int spa_ld_verify_pool_data(spa_t *spa) { int error = 0; vdev_t *rvd = spa->spa_root_vdev; /* * We've successfully opened the pool, verify that we're ready * to start pushing transactions. */ if (spa->spa_load_state != SPA_LOAD_TRYIMPORT) { error = spa_load_verify(spa); if (error != 0) { spa_load_failed(spa, "spa_load_verify failed " "[error=%d]", error); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } } return (0); } static void spa_ld_claim_log_blocks(spa_t *spa) { dmu_tx_t *tx; dsl_pool_t *dp = spa_get_dsl(spa); /* * Claim log blocks that haven't been committed yet. * This must all happen in a single txg. * Note: spa_claim_max_txg is updated by spa_claim_notify(), * invoked from zil_claim_log_block()'s i/o done callback. * Price of rollback is that we abandon the log. */ spa->spa_claiming = B_TRUE; tx = dmu_tx_create_assigned(dp, spa_first_txg(spa)); (void) dmu_objset_find_dp(dp, dp->dp_root_dir_obj, zil_claim, tx, DS_FIND_CHILDREN); dmu_tx_commit(tx); spa->spa_claiming = B_FALSE; spa_set_log_state(spa, SPA_LOG_GOOD); } static void spa_ld_check_for_config_update(spa_t *spa, uint64_t config_cache_txg, boolean_t update_config_cache) { vdev_t *rvd = spa->spa_root_vdev; int need_update = B_FALSE; /* * If the config cache is stale, or we have uninitialized * metaslabs (see spa_vdev_add()), then update the config. * * If this is a verbatim import, trust the current * in-core spa_config and update the disk labels. */ if (update_config_cache || config_cache_txg != spa->spa_config_txg || spa->spa_load_state == SPA_LOAD_IMPORT || spa->spa_load_state == SPA_LOAD_RECOVER || (spa->spa_import_flags & ZFS_IMPORT_VERBATIM)) need_update = B_TRUE; for (int c = 0; c < rvd->vdev_children; c++) if (rvd->vdev_child[c]->vdev_ms_array == 0) need_update = B_TRUE; /* * Update the config cache asynchronously in case we're the * root pool, in which case the config cache isn't writable yet. */ if (need_update) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } static void spa_ld_prepare_for_reload(spa_t *spa) { spa_mode_t mode = spa->spa_mode; int async_suspended = spa->spa_async_suspended; spa_unload(spa); spa_deactivate(spa); spa_activate(spa, mode); /* * We save the value of spa_async_suspended as it gets reset to 0 by * spa_unload(). We want to restore it back to the original value before * returning as we might be calling spa_async_resume() later. */ spa->spa_async_suspended = async_suspended; } static int spa_ld_read_checkpoint_txg(spa_t *spa) { uberblock_t checkpoint; int error = 0; ASSERT0(spa->spa_checkpoint_txg); ASSERT(MUTEX_HELD(&spa_namespace_lock)); error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t), sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint); if (error == ENOENT) return (0); if (error != 0) return (error); ASSERT3U(checkpoint.ub_txg, !=, 0); ASSERT3U(checkpoint.ub_checkpoint_txg, !=, 0); ASSERT3U(checkpoint.ub_timestamp, !=, 0); spa->spa_checkpoint_txg = checkpoint.ub_txg; spa->spa_checkpoint_info.sci_timestamp = checkpoint.ub_timestamp; return (0); } static int spa_ld_mos_init(spa_t *spa, spa_import_type_t type) { int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE); /* * Never trust the config that is provided unless we are assembling * a pool following a split. * This means don't trust blkptrs and the vdev tree in general. This * also effectively puts the spa in read-only mode since * spa_writeable() checks for spa_trust_config to be true. * We will later load a trusted config from the MOS. */ if (type != SPA_IMPORT_ASSEMBLE) spa->spa_trust_config = B_FALSE; /* * Parse the config provided to create a vdev tree. */ error = spa_ld_parse_config(spa, type); if (error != 0) return (error); spa_import_progress_add(spa); /* * Now that we have the vdev tree, try to open each vdev. This involves * opening the underlying physical device, retrieving its geometry and * probing the vdev with a dummy I/O. The state of each vdev will be set * based on the success of those operations. After this we'll be ready * to read from the vdevs. */ error = spa_ld_open_vdevs(spa); if (error != 0) return (error); /* * Read the label of each vdev and make sure that the GUIDs stored * there match the GUIDs in the config provided. * If we're assembling a new pool that's been split off from an * existing pool, the labels haven't yet been updated so we skip * validation for now. */ if (type != SPA_IMPORT_ASSEMBLE) { error = spa_ld_validate_vdevs(spa); if (error != 0) return (error); } /* * Read all vdev labels to find the best uberblock (i.e. latest, * unless spa_load_max_txg is set) and store it in spa_uberblock. We * get the list of features required to read blkptrs in the MOS from * the vdev label with the best uberblock and verify that our version * of zfs supports them all. */ error = spa_ld_select_uberblock(spa, type); if (error != 0) return (error); /* * Pass that uberblock to the dsl_pool layer which will open the root * blkptr. This blkptr points to the latest version of the MOS and will * allow us to read its contents. */ error = spa_ld_open_rootbp(spa); if (error != 0) return (error); return (0); } static int spa_ld_checkpoint_rewind(spa_t *spa) { uberblock_t checkpoint; int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT); error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t), sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint); if (error != 0) { spa_load_failed(spa, "unable to retrieve checkpointed " "uberblock from the MOS config [error=%d]", error); if (error == ENOENT) error = ZFS_ERR_NO_CHECKPOINT; return (error); } ASSERT3U(checkpoint.ub_txg, <, spa->spa_uberblock.ub_txg); ASSERT3U(checkpoint.ub_txg, ==, checkpoint.ub_checkpoint_txg); /* * We need to update the txg and timestamp of the checkpointed * uberblock to be higher than the latest one. This ensures that * the checkpointed uberblock is selected if we were to close and * reopen the pool right after we've written it in the vdev labels. * (also see block comment in vdev_uberblock_compare) */ checkpoint.ub_txg = spa->spa_uberblock.ub_txg + 1; checkpoint.ub_timestamp = gethrestime_sec(); /* * Set current uberblock to be the checkpointed uberblock. */ spa->spa_uberblock = checkpoint; /* * If we are doing a normal rewind, then the pool is open for * writing and we sync the "updated" checkpointed uberblock to * disk. Once this is done, we've basically rewound the whole * pool and there is no way back. * * There are cases when we don't want to attempt and sync the * checkpointed uberblock to disk because we are opening a * pool as read-only. Specifically, verifying the checkpointed * state with zdb, and importing the checkpointed state to get * a "preview" of its content. */ if (spa_writeable(spa)) { vdev_t *rvd = spa->spa_root_vdev; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL }; int svdcount = 0; int children = rvd->vdev_children; int c0 = random_in_range(children); for (int c = 0; c < children; c++) { vdev_t *vd = rvd->vdev_child[(c0 + c) % children]; /* Stop when revisiting the first vdev */ if (c > 0 && svd[0] == vd) break; if (vd->vdev_ms_array == 0 || vd->vdev_islog || !vdev_is_concrete(vd)) continue; svd[svdcount++] = vd; if (svdcount == SPA_SYNC_MIN_VDEVS) break; } error = vdev_config_sync(svd, svdcount, spa->spa_first_txg); if (error == 0) spa->spa_last_synced_guid = rvd->vdev_guid; spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_load_failed(spa, "failed to write checkpointed " "uberblock to the vdev labels [error=%d]", error); return (error); } } return (0); } static int spa_ld_mos_with_trusted_config(spa_t *spa, spa_import_type_t type, boolean_t *update_config_cache) { int error; /* * Parse the config for pool, open and validate vdevs, * select an uberblock, and use that uberblock to open * the MOS. */ error = spa_ld_mos_init(spa, type); if (error != 0) return (error); /* * Retrieve the trusted config stored in the MOS and use it to create * a new, exact version of the vdev tree, then reopen all vdevs. */ error = spa_ld_trusted_config(spa, type, B_FALSE); if (error == EAGAIN) { if (update_config_cache != NULL) *update_config_cache = B_TRUE; /* * Redo the loading process with the trusted config if it is * too different from the untrusted config. */ spa_ld_prepare_for_reload(spa); spa_load_note(spa, "RELOADING"); error = spa_ld_mos_init(spa, type); if (error != 0) return (error); error = spa_ld_trusted_config(spa, type, B_TRUE); if (error != 0) return (error); } else if (error != 0) { return (error); } return (0); } /* * Load an existing storage pool, using the config provided. This config * describes which vdevs are part of the pool and is later validated against * partial configs present in each vdev's label and an entire copy of the * config stored in the MOS. */ static int spa_load_impl(spa_t *spa, spa_import_type_t type, const char **ereport) { int error = 0; boolean_t missing_feat_write = B_FALSE; boolean_t checkpoint_rewind = (spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT); boolean_t update_config_cache = B_FALSE; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE); spa_load_note(spa, "LOADING"); error = spa_ld_mos_with_trusted_config(spa, type, &update_config_cache); if (error != 0) return (error); /* * If we are rewinding to the checkpoint then we need to repeat * everything we've done so far in this function but this time * selecting the checkpointed uberblock and using that to open * the MOS. */ if (checkpoint_rewind) { /* * If we are rewinding to the checkpoint update config cache * anyway. */ update_config_cache = B_TRUE; /* * Extract the checkpointed uberblock from the current MOS * and use this as the pool's uberblock from now on. If the * pool is imported as writeable we also write the checkpoint * uberblock to the labels, making the rewind permanent. */ error = spa_ld_checkpoint_rewind(spa); if (error != 0) return (error); /* * Redo the loading process again with the * checkpointed uberblock. */ spa_ld_prepare_for_reload(spa); spa_load_note(spa, "LOADING checkpointed uberblock"); error = spa_ld_mos_with_trusted_config(spa, type, NULL); if (error != 0) return (error); } /* * Retrieve the checkpoint txg if the pool has a checkpoint. */ error = spa_ld_read_checkpoint_txg(spa); if (error != 0) return (error); /* * Retrieve the mapping of indirect vdevs. Those vdevs were removed * from the pool and their contents were re-mapped to other vdevs. Note * that everything that we read before this step must have been * rewritten on concrete vdevs after the last device removal was * initiated. Otherwise we could be reading from indirect vdevs before * we have loaded their mappings. */ error = spa_ld_open_indirect_vdev_metadata(spa); if (error != 0) return (error); /* * Retrieve the full list of active features from the MOS and check if * they are all supported. */ error = spa_ld_check_features(spa, &missing_feat_write); if (error != 0) return (error); /* * Load several special directories from the MOS needed by the dsl_pool * layer. */ error = spa_ld_load_special_directories(spa); if (error != 0) return (error); /* * Retrieve pool properties from the MOS. */ error = spa_ld_get_props(spa); if (error != 0) return (error); /* * Retrieve the list of auxiliary devices - cache devices and spares - * and open them. */ error = spa_ld_open_aux_vdevs(spa, type); if (error != 0) return (error); /* * Load the metadata for all vdevs. Also check if unopenable devices * should be autoreplaced. */ error = spa_ld_load_vdev_metadata(spa); if (error != 0) return (error); error = spa_ld_load_dedup_tables(spa); if (error != 0) return (error); /* * Verify the logs now to make sure we don't have any unexpected errors * when we claim log blocks later. */ error = spa_ld_verify_logs(spa, type, ereport); if (error != 0) return (error); if (missing_feat_write) { ASSERT(spa->spa_load_state == SPA_LOAD_TRYIMPORT); /* * At this point, we know that we can open the pool in * read-only mode but not read-write mode. We now have enough * information and can return to userland. */ return (spa_vdev_err(spa->spa_root_vdev, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * Traverse the last txgs to make sure the pool was left off in a safe * state. When performing an extreme rewind, we verify the whole pool, * which can take a very long time. */ error = spa_ld_verify_pool_data(spa); if (error != 0) return (error); /* * Calculate the deflated space for the pool. This must be done before * we write anything to the pool because we'd need to update the space * accounting using the deflated sizes. */ spa_update_dspace(spa); /* * We have now retrieved all the information we needed to open the * pool. If we are importing the pool in read-write mode, a few * additional steps must be performed to finish the import. */ if (spa_writeable(spa) && (spa->spa_load_state == SPA_LOAD_RECOVER || spa->spa_load_max_txg == UINT64_MAX)) { uint64_t config_cache_txg = spa->spa_config_txg; ASSERT(spa->spa_load_state != SPA_LOAD_TRYIMPORT); /* * In case of a checkpoint rewind, log the original txg * of the checkpointed uberblock. */ if (checkpoint_rewind) { spa_history_log_internal(spa, "checkpoint rewind", NULL, "rewound state to txg=%llu", (u_longlong_t)spa->spa_uberblock.ub_checkpoint_txg); } /* * Traverse the ZIL and claim all blocks. */ spa_ld_claim_log_blocks(spa); /* * Kick-off the syncing thread. */ spa->spa_sync_on = B_TRUE; txg_sync_start(spa->spa_dsl_pool); mmp_thread_start(spa); /* * Wait for all claims to sync. We sync up to the highest * claimed log block birth time so that claimed log blocks * don't appear to be from the future. spa_claim_max_txg * will have been set for us by ZIL traversal operations * performed above. */ txg_wait_synced(spa->spa_dsl_pool, spa->spa_claim_max_txg); /* * Check if we need to request an update of the config. On the * next sync, we would update the config stored in vdev labels * and the cachefile (by default /etc/zfs/zpool.cache). */ spa_ld_check_for_config_update(spa, config_cache_txg, update_config_cache); /* * Check if a rebuild was in progress and if so resume it. * Then check all DTLs to see if anything needs resilvering. * The resilver will be deferred if a rebuild was started. */ if (vdev_rebuild_active(spa->spa_root_vdev)) { vdev_rebuild_restart(spa); } else if (!dsl_scan_resilvering(spa->spa_dsl_pool) && vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) { spa_async_request(spa, SPA_ASYNC_RESILVER); } /* * Log the fact that we booted up (so that we can detect if * we rebooted in the middle of an operation). */ spa_history_log_version(spa, "open", NULL); spa_restart_removal(spa); spa_spawn_aux_threads(spa); /* * Delete any inconsistent datasets. * * Note: * Since we may be issuing deletes for clones here, * we make sure to do so after we've spawned all the * auxiliary threads above (from which the livelist * deletion zthr is part of). */ (void) dmu_objset_find(spa_name(spa), dsl_destroy_inconsistent, NULL, DS_FIND_CHILDREN); /* * Clean up any stale temporary dataset userrefs. */ dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); vdev_initialize_restart(spa->spa_root_vdev); vdev_trim_restart(spa->spa_root_vdev); vdev_autotrim_restart(spa); spa_config_exit(spa, SCL_CONFIG, FTAG); } spa_import_progress_remove(spa_guid(spa)); spa_async_request(spa, SPA_ASYNC_L2CACHE_REBUILD); spa_load_note(spa, "LOADED"); return (0); } static int spa_load_retry(spa_t *spa, spa_load_state_t state) { spa_mode_t mode = spa->spa_mode; spa_unload(spa); spa_deactivate(spa); spa->spa_load_max_txg = spa->spa_uberblock.ub_txg - 1; spa_activate(spa, mode); spa_async_suspend(spa); spa_load_note(spa, "spa_load_retry: rewind, max txg: %llu", (u_longlong_t)spa->spa_load_max_txg); return (spa_load(spa, state, SPA_IMPORT_EXISTING)); } /* * If spa_load() fails this function will try loading prior txg's. If * 'state' is SPA_LOAD_RECOVER and one of these loads succeeds the pool * will be rewound to that txg. If 'state' is not SPA_LOAD_RECOVER this * function will not rewind the pool and will return the same error as * spa_load(). */ static int spa_load_best(spa_t *spa, spa_load_state_t state, uint64_t max_request, int rewind_flags) { nvlist_t *loadinfo = NULL; nvlist_t *config = NULL; int load_error, rewind_error; uint64_t safe_rewind_txg; uint64_t min_txg; if (spa->spa_load_txg && state == SPA_LOAD_RECOVER) { spa->spa_load_max_txg = spa->spa_load_txg; spa_set_log_state(spa, SPA_LOG_CLEAR); } else { spa->spa_load_max_txg = max_request; if (max_request != UINT64_MAX) spa->spa_extreme_rewind = B_TRUE; } load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING); if (load_error == 0) return (0); if (load_error == ZFS_ERR_NO_CHECKPOINT) { /* * When attempting checkpoint-rewind on a pool with no * checkpoint, we should not attempt to load uberblocks * from previous txgs when spa_load fails. */ ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT); spa_import_progress_remove(spa_guid(spa)); return (load_error); } if (spa->spa_root_vdev != NULL) config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); spa->spa_last_ubsync_txg = spa->spa_uberblock.ub_txg; spa->spa_last_ubsync_txg_ts = spa->spa_uberblock.ub_timestamp; if (rewind_flags & ZPOOL_NEVER_REWIND) { nvlist_free(config); spa_import_progress_remove(spa_guid(spa)); return (load_error); } if (state == SPA_LOAD_RECOVER) { /* Price of rolling back is discarding txgs, including log */ spa_set_log_state(spa, SPA_LOG_CLEAR); } else { /* * If we aren't rolling back save the load info from our first * import attempt so that we can restore it after attempting * to rewind. */ loadinfo = spa->spa_load_info; spa->spa_load_info = fnvlist_alloc(); } spa->spa_load_max_txg = spa->spa_last_ubsync_txg; safe_rewind_txg = spa->spa_last_ubsync_txg - TXG_DEFER_SIZE; min_txg = (rewind_flags & ZPOOL_EXTREME_REWIND) ? TXG_INITIAL : safe_rewind_txg; /* * Continue as long as we're finding errors, we're still within * the acceptable rewind range, and we're still finding uberblocks */ while (rewind_error && spa->spa_uberblock.ub_txg >= min_txg && spa->spa_uberblock.ub_txg <= spa->spa_load_max_txg) { if (spa->spa_load_max_txg < safe_rewind_txg) spa->spa_extreme_rewind = B_TRUE; rewind_error = spa_load_retry(spa, state); } spa->spa_extreme_rewind = B_FALSE; spa->spa_load_max_txg = UINT64_MAX; if (config && (rewind_error || state != SPA_LOAD_RECOVER)) spa_config_set(spa, config); else nvlist_free(config); if (state == SPA_LOAD_RECOVER) { ASSERT3P(loadinfo, ==, NULL); spa_import_progress_remove(spa_guid(spa)); return (rewind_error); } else { /* Store the rewind info as part of the initial load info */ fnvlist_add_nvlist(loadinfo, ZPOOL_CONFIG_REWIND_INFO, spa->spa_load_info); /* Restore the initial load info */ fnvlist_free(spa->spa_load_info); spa->spa_load_info = loadinfo; spa_import_progress_remove(spa_guid(spa)); return (load_error); } } /* * Pool Open/Import * * The import case is identical to an open except that the configuration is sent * down from userland, instead of grabbed from the configuration cache. For the * case of an open, the pool configuration will exist in the * POOL_STATE_UNINITIALIZED state. * * The stats information (gen/count/ustats) is used to gather vdev statistics at * the same time open the pool, without having to keep around the spa_t in some * ambiguous state. */ static int spa_open_common(const char *pool, spa_t **spapp, const void *tag, nvlist_t *nvpolicy, nvlist_t **config) { spa_t *spa; spa_load_state_t state = SPA_LOAD_OPEN; int error; int locked = B_FALSE; int firstopen = B_FALSE; *spapp = NULL; /* * As disgusting as this is, we need to support recursive calls to this * function because dsl_dir_open() is called during spa_load(), and ends * up calling spa_open() again. The real fix is to figure out how to * avoid dsl_dir_open() calling this in the first place. */ if (MUTEX_NOT_HELD(&spa_namespace_lock)) { mutex_enter(&spa_namespace_lock); locked = B_TRUE; } if ((spa = spa_lookup(pool)) == NULL) { if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (spa->spa_state == POOL_STATE_UNINITIALIZED) { zpool_load_policy_t policy; firstopen = B_TRUE; zpool_get_load_policy(nvpolicy ? nvpolicy : spa->spa_config, &policy); if (policy.zlp_rewind & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; spa_activate(spa, spa_mode_global); if (state != SPA_LOAD_RECOVER) spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE; zfs_dbgmsg("spa_open_common: opening %s", pool); error = spa_load_best(spa, state, policy.zlp_txg, policy.zlp_rewind); if (error == EBADF) { /* * If vdev_validate() returns failure (indicated by * EBADF), it indicates that one of the vdevs indicates * that the pool has been exported or destroyed. If * this is the case, the config cache is out of sync and * we should remove the pool from the namespace. */ spa_unload(spa); spa_deactivate(spa); - spa_write_cachefile(spa, B_TRUE, B_TRUE); + spa_write_cachefile(spa, B_TRUE, B_TRUE, B_FALSE); spa_remove(spa); if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (error) { /* * We can't open the pool, but we still have useful * information: the state of each vdev after the * attempted vdev_open(). Return this to the user. */ if (config != NULL && spa->spa_config) { *config = fnvlist_dup(spa->spa_config); fnvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info); } spa_unload(spa); spa_deactivate(spa); spa->spa_last_open_failed = error; if (locked) mutex_exit(&spa_namespace_lock); *spapp = NULL; return (error); } } spa_open_ref(spa, tag); if (config != NULL) *config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); /* * If we've recovered the pool, pass back any information we * gathered while doing the load. */ if (state == SPA_LOAD_RECOVER) { fnvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info); } if (locked) { spa->spa_last_open_failed = 0; spa->spa_last_ubsync_txg = 0; spa->spa_load_txg = 0; mutex_exit(&spa_namespace_lock); } if (firstopen) zvol_create_minors_recursive(spa_name(spa)); *spapp = spa; return (0); } int spa_open_rewind(const char *name, spa_t **spapp, const void *tag, nvlist_t *policy, nvlist_t **config) { return (spa_open_common(name, spapp, tag, policy, config)); } int spa_open(const char *name, spa_t **spapp, const void *tag) { return (spa_open_common(name, spapp, tag, NULL, NULL)); } /* * Lookup the given spa_t, incrementing the inject count in the process, * preventing it from being exported or destroyed. */ spa_t * spa_inject_addref(char *name) { spa_t *spa; mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(name)) == NULL) { mutex_exit(&spa_namespace_lock); return (NULL); } spa->spa_inject_ref++; mutex_exit(&spa_namespace_lock); return (spa); } void spa_inject_delref(spa_t *spa) { mutex_enter(&spa_namespace_lock); spa->spa_inject_ref--; mutex_exit(&spa_namespace_lock); } /* * Add spares device information to the nvlist. */ static void spa_add_spares(spa_t *spa, nvlist_t *config) { nvlist_t **spares; uint_t i, nspares; nvlist_t *nvroot; uint64_t guid; vdev_stat_t *vs; uint_t vsc; uint64_t pool; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_spares.sav_count == 0) return; nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE); VERIFY0(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares)); if (nspares != 0) { fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares, nspares); VERIFY0(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares)); /* * Go through and find any spares which have since been * repurposed as an active spare. If this is the case, update * their status appropriately. */ for (i = 0; i < nspares; i++) { guid = fnvlist_lookup_uint64(spares[i], ZPOOL_CONFIG_GUID); if (spa_spare_exists(guid, &pool, NULL) && pool != 0ULL) { VERIFY0(nvlist_lookup_uint64_array(spares[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc)); vs->vs_state = VDEV_STATE_CANT_OPEN; vs->vs_aux = VDEV_AUX_SPARED; } } } } /* * Add l2cache device information to the nvlist, including vdev stats. */ static void spa_add_l2cache(spa_t *spa, nvlist_t *config) { nvlist_t **l2cache; uint_t i, j, nl2cache; nvlist_t *nvroot; uint64_t guid; vdev_t *vd; vdev_stat_t *vs; uint_t vsc; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_l2cache.sav_count == 0) return; nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE); VERIFY0(nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache)); if (nl2cache != 0) { fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache, nl2cache); VERIFY0(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache)); /* * Update level 2 cache device stats. */ for (i = 0; i < nl2cache; i++) { guid = fnvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID); vd = NULL; for (j = 0; j < spa->spa_l2cache.sav_count; j++) { if (guid == spa->spa_l2cache.sav_vdevs[j]->vdev_guid) { vd = spa->spa_l2cache.sav_vdevs[j]; break; } } ASSERT(vd != NULL); VERIFY0(nvlist_lookup_uint64_array(l2cache[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc)); vdev_get_stats(vd, vs); vdev_config_generate_stats(vd, l2cache[i]); } } } static void spa_feature_stats_from_disk(spa_t *spa, nvlist_t *features) { zap_cursor_t zc; zap_attribute_t za; if (spa->spa_feat_for_read_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_read_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY0(nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } if (spa->spa_feat_for_write_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_write_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY0(nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } } static void spa_feature_stats_from_cache(spa_t *spa, nvlist_t *features) { int i; for (i = 0; i < SPA_FEATURES; i++) { zfeature_info_t feature = spa_feature_table[i]; uint64_t refcount; if (feature_get_refcount(spa, &feature, &refcount) != 0) continue; VERIFY0(nvlist_add_uint64(features, feature.fi_guid, refcount)); } } /* * Store a list of pool features and their reference counts in the * config. * * The first time this is called on a spa, allocate a new nvlist, fetch * the pool features and reference counts from disk, then save the list * in the spa. In subsequent calls on the same spa use the saved nvlist * and refresh its values from the cached reference counts. This * ensures we don't block here on I/O on a suspended pool so 'zpool * clear' can resume the pool. */ static void spa_add_feature_stats(spa_t *spa, nvlist_t *config) { nvlist_t *features; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); mutex_enter(&spa->spa_feat_stats_lock); features = spa->spa_feat_stats; if (features != NULL) { spa_feature_stats_from_cache(spa, features); } else { VERIFY0(nvlist_alloc(&features, NV_UNIQUE_NAME, KM_SLEEP)); spa->spa_feat_stats = features; spa_feature_stats_from_disk(spa, features); } VERIFY0(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURE_STATS, features)); mutex_exit(&spa->spa_feat_stats_lock); } int spa_get_stats(const char *name, nvlist_t **config, char *altroot, size_t buflen) { int error; spa_t *spa; *config = NULL; error = spa_open_common(name, &spa, FTAG, NULL, config); if (spa != NULL) { /* * This still leaves a window of inconsistency where the spares * or l2cache devices could change and the config would be * self-inconsistent. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); if (*config != NULL) { uint64_t loadtimes[2]; loadtimes[0] = spa->spa_loaded_ts.tv_sec; loadtimes[1] = spa->spa_loaded_ts.tv_nsec; fnvlist_add_uint64_array(*config, ZPOOL_CONFIG_LOADED_TIME, loadtimes, 2); fnvlist_add_uint64(*config, ZPOOL_CONFIG_ERRCOUNT, spa_get_errlog_size(spa)); if (spa_suspended(spa)) { fnvlist_add_uint64(*config, ZPOOL_CONFIG_SUSPENDED, spa->spa_failmode); fnvlist_add_uint64(*config, ZPOOL_CONFIG_SUSPENDED_REASON, spa->spa_suspended); } spa_add_spares(spa, *config); spa_add_l2cache(spa, *config); spa_add_feature_stats(spa, *config); } } /* * We want to get the alternate root even for faulted pools, so we cheat * and call spa_lookup() directly. */ if (altroot) { if (spa == NULL) { mutex_enter(&spa_namespace_lock); spa = spa_lookup(name); if (spa) spa_altroot(spa, altroot, buflen); else altroot[0] = '\0'; spa = NULL; mutex_exit(&spa_namespace_lock); } else { spa_altroot(spa, altroot, buflen); } } if (spa != NULL) { spa_config_exit(spa, SCL_CONFIG, FTAG); spa_close(spa, FTAG); } return (error); } /* * Validate that the auxiliary device array is well formed. We must have an * array of nvlists, each which describes a valid leaf vdev. If this is an * import (mode is VDEV_ALLOC_SPARE), then we allow corrupted spares to be * specified, as long as they are well-formed. */ static int spa_validate_aux_devs(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode, spa_aux_vdev_t *sav, const char *config, uint64_t version, vdev_labeltype_t label) { nvlist_t **dev; uint_t i, ndev; vdev_t *vd; int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * It's acceptable to have no devs specified. */ if (nvlist_lookup_nvlist_array(nvroot, config, &dev, &ndev) != 0) return (0); if (ndev == 0) return (SET_ERROR(EINVAL)); /* * Make sure the pool is formatted with a version that supports this * device type. */ if (spa_version(spa) < version) return (SET_ERROR(ENOTSUP)); /* * Set the pending device list so we correctly handle device in-use * checking. */ sav->sav_pending = dev; sav->sav_npending = ndev; for (i = 0; i < ndev; i++) { if ((error = spa_config_parse(spa, &vd, dev[i], NULL, 0, mode)) != 0) goto out; if (!vd->vdev_ops->vdev_op_leaf) { vdev_free(vd); error = SET_ERROR(EINVAL); goto out; } vd->vdev_top = vd; if ((error = vdev_open(vd)) == 0 && (error = vdev_label_init(vd, crtxg, label)) == 0) { fnvlist_add_uint64(dev[i], ZPOOL_CONFIG_GUID, vd->vdev_guid); } vdev_free(vd); if (error && (mode != VDEV_ALLOC_SPARE && mode != VDEV_ALLOC_L2CACHE)) goto out; else error = 0; } out: sav->sav_pending = NULL; sav->sav_npending = 0; return (error); } static int spa_validate_aux(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode) { int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if ((error = spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_spares, ZPOOL_CONFIG_SPARES, SPA_VERSION_SPARES, VDEV_LABEL_SPARE)) != 0) { return (error); } return (spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_l2cache, ZPOOL_CONFIG_L2CACHE, SPA_VERSION_L2CACHE, VDEV_LABEL_L2CACHE)); } static void spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs, const char *config) { int i; if (sav->sav_config != NULL) { nvlist_t **olddevs; uint_t oldndevs; nvlist_t **newdevs; /* * Generate new dev list by concatenating with the * current dev list. */ VERIFY0(nvlist_lookup_nvlist_array(sav->sav_config, config, &olddevs, &oldndevs)); newdevs = kmem_alloc(sizeof (void *) * (ndevs + oldndevs), KM_SLEEP); for (i = 0; i < oldndevs; i++) newdevs[i] = fnvlist_dup(olddevs[i]); for (i = 0; i < ndevs; i++) newdevs[i + oldndevs] = fnvlist_dup(devs[i]); fnvlist_remove(sav->sav_config, config); fnvlist_add_nvlist_array(sav->sav_config, config, (const nvlist_t * const *)newdevs, ndevs + oldndevs); for (i = 0; i < oldndevs + ndevs; i++) nvlist_free(newdevs[i]); kmem_free(newdevs, (oldndevs + ndevs) * sizeof (void *)); } else { /* * Generate a new dev list. */ sav->sav_config = fnvlist_alloc(); fnvlist_add_nvlist_array(sav->sav_config, config, (const nvlist_t * const *)devs, ndevs); } } /* * Stop and drop level 2 ARC devices */ void spa_l2cache_drop(spa_t *spa) { vdev_t *vd; int i; spa_aux_vdev_t *sav = &spa->spa_l2cache; for (i = 0; i < sav->sav_count; i++) { uint64_t pool; vd = sav->sav_vdevs[i]; ASSERT(vd != NULL); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); } } /* * Verify encryption parameters for spa creation. If we are encrypting, we must * have the encryption feature flag enabled. */ static int spa_create_check_encryption_params(dsl_crypto_params_t *dcp, boolean_t has_encryption) { if (dcp->cp_crypt != ZIO_CRYPT_OFF && dcp->cp_crypt != ZIO_CRYPT_INHERIT && !has_encryption) return (SET_ERROR(ENOTSUP)); return (dmu_objset_create_crypt_check(NULL, dcp, NULL)); } /* * Pool Creation */ int spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *zplprops, dsl_crypto_params_t *dcp) { spa_t *spa; char *altroot = NULL; vdev_t *rvd; dsl_pool_t *dp; dmu_tx_t *tx; int error = 0; uint64_t txg = TXG_INITIAL; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; uint64_t version, obj, ndraid = 0; boolean_t has_features; boolean_t has_encryption; boolean_t has_allocclass; spa_feature_t feat; char *feat_name; char *poolname; nvlist_t *nvl; if (props == NULL || nvlist_lookup_string(props, "tname", &poolname) != 0) poolname = (char *)pool; /* * If this pool already exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(poolname) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Allocate a new spa_t structure. */ nvl = fnvlist_alloc(); fnvlist_add_string(nvl, ZPOOL_CONFIG_POOL_NAME, pool); (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); spa = spa_add(poolname, nvl, altroot); fnvlist_free(nvl); spa_activate(spa, spa_mode_global); if (props && (error = spa_prop_validate(spa, props))) { spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Temporary pool names should never be written to disk. */ if (poolname != pool) spa->spa_import_flags |= ZFS_IMPORT_TEMP_NAME; has_features = B_FALSE; has_encryption = B_FALSE; has_allocclass = B_FALSE; for (nvpair_t *elem = nvlist_next_nvpair(props, NULL); elem != NULL; elem = nvlist_next_nvpair(props, elem)) { if (zpool_prop_feature(nvpair_name(elem))) { has_features = B_TRUE; feat_name = strchr(nvpair_name(elem), '@') + 1; VERIFY0(zfeature_lookup_name(feat_name, &feat)); if (feat == SPA_FEATURE_ENCRYPTION) has_encryption = B_TRUE; if (feat == SPA_FEATURE_ALLOCATION_CLASSES) has_allocclass = B_TRUE; } } /* verify encryption params, if they were provided */ if (dcp != NULL) { error = spa_create_check_encryption_params(dcp, has_encryption); if (error != 0) { spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } } if (!has_allocclass && zfs_special_devs(nvroot, NULL)) { spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (ENOTSUP); } if (has_features || nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version) != 0) { version = SPA_VERSION; } ASSERT(SPA_VERSION_IS_SUPPORTED(version)); spa->spa_first_txg = txg; spa->spa_uberblock.ub_txg = txg - 1; spa->spa_uberblock.ub_version = version; spa->spa_ubsync = spa->spa_uberblock; spa->spa_load_state = SPA_LOAD_CREATE; spa->spa_removing_phys.sr_state = DSS_NONE; spa->spa_removing_phys.sr_removing_vdev = -1; spa->spa_removing_phys.sr_prev_indirect_vdev = -1; spa->spa_indirect_vdevs_loaded = B_TRUE; /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *), KM_SLEEP); for (int i = 0; i < max_ncpus; i++) { spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); } /* * Create the root vdev. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_ADD); ASSERT(error != 0 || rvd != NULL); ASSERT(error != 0 || spa->spa_root_vdev == rvd); if (error == 0 && !zfs_allocatable_devs(nvroot)) error = SET_ERROR(EINVAL); if (error == 0 && (error = vdev_create(rvd, txg, B_FALSE)) == 0 && (error = vdev_draid_spare_create(nvroot, rvd, &ndraid, 0)) == 0 && (error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) == 0) { /* * instantiate the metaslab groups (this will dirty the vdevs) * we can no longer error exit past this point */ for (int c = 0; error == 0 && c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; vdev_metaslab_set_size(vd); vdev_expand(vd, txg); } } spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Get the list of spares, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { spa->spa_spares.sav_config = fnvlist_alloc(); fnvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares, nspares); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } /* * Get the list of level 2 cache devices, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { VERIFY0(nvlist_alloc(&spa->spa_l2cache.sav_config, NV_UNIQUE_NAME, KM_SLEEP)); fnvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache, nl2cache); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } spa->spa_is_initializing = B_TRUE; spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, dcp, txg); spa->spa_is_initializing = B_FALSE; /* * Create DDTs (dedup tables). */ ddt_create(spa); spa_update_dspace(spa); tx = dmu_tx_create_assigned(dp, txg); /* * Create the pool's history object. */ if (version >= SPA_VERSION_ZPOOL_HISTORY && !spa->spa_history) spa_history_create_obj(spa, tx); spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_CREATE); spa_history_log_version(spa, "create", tx); /* * Create the pool config object. */ spa->spa_config_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, SPA_CONFIG_BLOCKSIZE, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CONFIG, sizeof (uint64_t), 1, &spa->spa_config_object, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool config"); } if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION, sizeof (uint64_t), 1, &version, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool version"); } /* Newly created pools with the right version are always deflated. */ if (version >= SPA_VERSION_RAIDZ_DEFLATE) { spa->spa_deflate = TRUE; if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx) != 0) { cmn_err(CE_PANIC, "failed to add deflate"); } } /* * Create the deferred-free bpobj. Turn off compression * because sync-to-convergence takes longer if the blocksize * keeps changing. */ obj = bpobj_alloc(spa->spa_meta_objset, 1 << 14, tx); dmu_object_set_compress(spa->spa_meta_objset, obj, ZIO_COMPRESS_OFF, tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SYNC_BPOBJ, sizeof (uint64_t), 1, &obj, tx) != 0) { cmn_err(CE_PANIC, "failed to add bpobj"); } VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj)); /* * Generate some random noise for salted checksums to operate on. */ (void) random_get_pseudo_bytes(spa->spa_cksum_salt.zcs_bytes, sizeof (spa->spa_cksum_salt.zcs_bytes)); /* * Set pool properties. */ spa->spa_bootfs = zpool_prop_default_numeric(ZPOOL_PROP_BOOTFS); spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); spa->spa_failmode = zpool_prop_default_numeric(ZPOOL_PROP_FAILUREMODE); spa->spa_autoexpand = zpool_prop_default_numeric(ZPOOL_PROP_AUTOEXPAND); spa->spa_multihost = zpool_prop_default_numeric(ZPOOL_PROP_MULTIHOST); spa->spa_autotrim = zpool_prop_default_numeric(ZPOOL_PROP_AUTOTRIM); if (props != NULL) { spa_configfile_set(spa, props, B_FALSE); spa_sync_props(props, tx); } for (int i = 0; i < ndraid; i++) spa_feature_incr(spa, SPA_FEATURE_DRAID, tx); dmu_tx_commit(tx); spa->spa_sync_on = B_TRUE; txg_sync_start(dp); mmp_thread_start(spa); txg_wait_synced(dp, txg); spa_spawn_aux_threads(spa); - spa_write_cachefile(spa, B_FALSE, B_TRUE); + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_TRUE); /* * Don't count references from objsets that are already closed * and are making their way through the eviction process. */ spa_evicting_os_wait(spa); spa->spa_minref = zfs_refcount_count(&spa->spa_refcount); spa->spa_load_state = SPA_LOAD_NONE; spa_import_os(spa); mutex_exit(&spa_namespace_lock); return (0); } /* * Import a non-root pool into the system. */ int spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags) { spa_t *spa; char *altroot = NULL; spa_load_state_t state = SPA_LOAD_IMPORT; zpool_load_policy_t policy; spa_mode_t mode = spa_mode_global; uint64_t readonly = B_FALSE; int error; nvlist_t *nvroot; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; /* * If a pool with this name exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(pool) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Create and initialize the spa structure. */ (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly); if (readonly) mode = SPA_MODE_READ; spa = spa_add(pool, config, altroot); spa->spa_import_flags = flags; /* * Verbatim import - Take a pool and insert it into the namespace * as if it had been loaded at boot. */ if (spa->spa_import_flags & ZFS_IMPORT_VERBATIM) { if (props != NULL) spa_configfile_set(spa, props, B_FALSE); - spa_write_cachefile(spa, B_FALSE, B_TRUE); + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_FALSE); spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT); zfs_dbgmsg("spa_import: verbatim import of %s", pool); mutex_exit(&spa_namespace_lock); return (0); } spa_activate(spa, mode); /* * Don't start async tasks until we know everything is healthy. */ spa_async_suspend(spa); zpool_get_load_policy(config, &policy); if (policy.zlp_rewind & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; spa->spa_config_source = SPA_CONFIG_SRC_TRYIMPORT; if (state != SPA_LOAD_RECOVER) { spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; zfs_dbgmsg("spa_import: importing %s", pool); } else { zfs_dbgmsg("spa_import: importing %s, max_txg=%lld " "(RECOVERY MODE)", pool, (longlong_t)policy.zlp_txg); } error = spa_load_best(spa, state, policy.zlp_txg, policy.zlp_rewind); /* * Propagate anything learned while loading the pool and pass it * back to caller (i.e. rewind info, missing devices, etc). */ fnvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Toss any existing sparelist, as it doesn't have any validity * anymore, and conflicts with spa_has_spare(). */ if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; spa_load_spares(spa); } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; spa_load_l2cache(spa); } nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE); spa_config_exit(spa, SCL_ALL, FTAG); if (props != NULL) spa_configfile_set(spa, props, B_FALSE); if (error != 0 || (props && spa_writeable(spa) && (error = spa_prop_set(spa, props)))) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } spa_async_resume(spa); /* * Override any spares and level 2 cache devices as specified by * the user, as these may have correct device names/devids, etc. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { if (spa->spa_spares.sav_config) fnvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES); else spa->spa_spares.sav_config = fnvlist_alloc(); fnvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares, nspares); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { if (spa->spa_l2cache.sav_config) fnvlist_remove(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE); else spa->spa_l2cache.sav_config = fnvlist_alloc(); fnvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache, nl2cache); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } /* * Check for any removed devices. */ if (spa->spa_autoreplace) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } if (spa_writeable(spa)) { /* * Update the config cache to include the newly-imported pool. */ spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); } /* * It's possible that the pool was expanded while it was exported. * We kick off an async task to handle this for us. */ spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); spa_history_log_version(spa, "import", NULL); spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT); mutex_exit(&spa_namespace_lock); zvol_create_minors_recursive(pool); spa_import_os(spa); return (0); } nvlist_t * spa_tryimport(nvlist_t *tryconfig) { nvlist_t *config = NULL; char *poolname, *cachefile; spa_t *spa; uint64_t state; int error; zpool_load_policy_t policy; if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname)) return (NULL); if (nvlist_lookup_uint64(tryconfig, ZPOOL_CONFIG_POOL_STATE, &state)) return (NULL); /* * Create and initialize the spa structure. */ mutex_enter(&spa_namespace_lock); spa = spa_add(TRYIMPORT_NAME, tryconfig, NULL); spa_activate(spa, SPA_MODE_READ); /* * Rewind pool if a max txg was provided. */ zpool_get_load_policy(spa->spa_config, &policy); if (policy.zlp_txg != UINT64_MAX) { spa->spa_load_max_txg = policy.zlp_txg; spa->spa_extreme_rewind = B_TRUE; zfs_dbgmsg("spa_tryimport: importing %s, max_txg=%lld", poolname, (longlong_t)policy.zlp_txg); } else { zfs_dbgmsg("spa_tryimport: importing %s", poolname); } if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_CACHEFILE, &cachefile) == 0) { zfs_dbgmsg("spa_tryimport: using cachefile '%s'", cachefile); spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE; } else { spa->spa_config_source = SPA_CONFIG_SRC_SCAN; } error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING); /* * If 'tryconfig' was at least parsable, return the current config. */ if (spa->spa_root_vdev != NULL) { config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, poolname); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, state); fnvlist_add_uint64(config, ZPOOL_CONFIG_TIMESTAMP, spa->spa_uberblock.ub_timestamp); fnvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info); fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata); /* * If the bootfs property exists on this pool then we * copy it out so that external consumers can tell which * pools are bootable. */ if ((!error || error == EEXIST) && spa->spa_bootfs) { char *tmpname = kmem_alloc(MAXPATHLEN, KM_SLEEP); /* * We have to play games with the name since the * pool was opened as TRYIMPORT_NAME. */ if (dsl_dsobj_to_dsname(spa_name(spa), spa->spa_bootfs, tmpname) == 0) { char *cp; char *dsname; dsname = kmem_alloc(MAXPATHLEN, KM_SLEEP); cp = strchr(tmpname, '/'); if (cp == NULL) { (void) strlcpy(dsname, tmpname, MAXPATHLEN); } else { (void) snprintf(dsname, MAXPATHLEN, "%s/%s", poolname, ++cp); } fnvlist_add_string(config, ZPOOL_CONFIG_BOOTFS, dsname); kmem_free(dsname, MAXPATHLEN); } kmem_free(tmpname, MAXPATHLEN); } /* * Add the list of hot spares and level 2 cache devices. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_add_spares(spa, config); spa_add_l2cache(spa, config); spa_config_exit(spa, SCL_CONFIG, FTAG); } spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (config); } /* * Pool export/destroy * * The act of destroying or exporting a pool is very simple. We make sure there * is no more pending I/O and any references to the pool are gone. Then, we * update the pool state and sync all the labels to disk, removing the * configuration from the cache afterwards. If the 'hardforce' flag is set, then * we don't sync the labels or remove the configuration cache. */ static int spa_export_common(const char *pool, int new_state, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { int error; spa_t *spa; if (oldconfig) *oldconfig = NULL; if (!(spa_mode_global & SPA_MODE_WRITE)) return (SET_ERROR(EROFS)); mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(pool)) == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (spa->spa_is_exporting) { /* the pool is being exported by another thread */ mutex_exit(&spa_namespace_lock); return (SET_ERROR(ZFS_ERR_EXPORT_IN_PROGRESS)); } spa->spa_is_exporting = B_TRUE; /* * Put a hold on the pool, drop the namespace lock, stop async tasks, * reacquire the namespace lock, and see if we can export. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); if (spa->spa_zvol_taskq) { zvol_remove_minors(spa, spa_name(spa), B_TRUE); taskq_wait(spa->spa_zvol_taskq); } mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); if (spa->spa_state == POOL_STATE_UNINITIALIZED) goto export_spa; /* * The pool will be in core if it's openable, in which case we can * modify its state. Objsets may be open only because they're dirty, * so we have to force it to sync before checking spa_refcnt. */ if (spa->spa_sync_on) { txg_wait_synced(spa->spa_dsl_pool, 0); spa_evicting_os_wait(spa); } /* * A pool cannot be exported or destroyed if there are active * references. If we are resetting a pool, allow references by * fault injection handlers. */ if (!spa_refcount_zero(spa) || (spa->spa_inject_ref != 0)) { error = SET_ERROR(EBUSY); goto fail; } if (spa->spa_sync_on) { vdev_t *rvd = spa->spa_root_vdev; /* * A pool cannot be exported if it has an active shared spare. * This is to prevent other pools stealing the active spare * from an exported pool. At user's own will, such pool can * be forcedly exported. */ if (!force && new_state == POOL_STATE_EXPORTED && spa_has_active_shared_spare(spa)) { error = SET_ERROR(EXDEV); goto fail; } /* * We're about to export or destroy this pool. Make sure * we stop all initialization and trim activity here before * we set the spa_final_txg. This will ensure that all * dirty data resulting from the initialization is * committed to disk before we unload the pool. */ vdev_initialize_stop_all(rvd, VDEV_INITIALIZE_ACTIVE); vdev_trim_stop_all(rvd, VDEV_TRIM_ACTIVE); vdev_autotrim_stop_all(spa); vdev_rebuild_stop_all(spa); /* * We want this to be reflected on every label, * so mark them all dirty. spa_unload() will do the * final sync that pushes these changes out. */ if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa->spa_state = new_state; vdev_config_dirty(rvd); spa_config_exit(spa, SCL_ALL, FTAG); } /* * If the log space map feature is enabled and the pool is * getting exported (but not destroyed), we want to spend some * time flushing as many metaslabs as we can in an attempt to * destroy log space maps and save import time. This has to be * done before we set the spa_final_txg, otherwise * spa_sync() -> spa_flush_metaslabs() may dirty the final TXGs. * spa_should_flush_logs_on_unload() should be called after * spa_state has been set to the new_state. */ if (spa_should_flush_logs_on_unload(spa)) spa_unload_log_sm_flush_all(spa); if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa->spa_final_txg = spa_last_synced_txg(spa) + TXG_DEFER_SIZE + 1; spa_config_exit(spa, SCL_ALL, FTAG); } } export_spa: spa_export_os(spa); if (new_state == POOL_STATE_DESTROYED) spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_DESTROY); else if (new_state == POOL_STATE_EXPORTED) spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_EXPORT); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } if (oldconfig && spa->spa_config) *oldconfig = fnvlist_dup(spa->spa_config); if (new_state != POOL_STATE_UNINITIALIZED) { if (!hardforce) - spa_write_cachefile(spa, B_TRUE, B_TRUE); + spa_write_cachefile(spa, B_TRUE, B_TRUE, B_FALSE); spa_remove(spa); } else { /* * If spa_remove() is not called for this spa_t and * there is any possibility that it can be reused, * we make sure to reset the exporting flag. */ spa->spa_is_exporting = B_FALSE; } mutex_exit(&spa_namespace_lock); return (0); fail: spa->spa_is_exporting = B_FALSE; spa_async_resume(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Destroy a storage pool. */ int spa_destroy(const char *pool) { return (spa_export_common(pool, POOL_STATE_DESTROYED, NULL, B_FALSE, B_FALSE)); } /* * Export a storage pool. */ int spa_export(const char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { return (spa_export_common(pool, POOL_STATE_EXPORTED, oldconfig, force, hardforce)); } /* * Similar to spa_export(), this unloads the spa_t without actually removing it * from the namespace in any way. */ int spa_reset(const char *pool) { return (spa_export_common(pool, POOL_STATE_UNINITIALIZED, NULL, B_FALSE, B_FALSE)); } /* * ========================================================================== * Device manipulation * ========================================================================== */ /* * This is called as a synctask to increment the draid feature flag */ static void spa_draid_feature_incr(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; int draid = (int)(uintptr_t)arg; for (int c = 0; c < draid; c++) spa_feature_incr(spa, SPA_FEATURE_DRAID, tx); } /* * Add a device to a storage pool. */ int spa_vdev_add(spa_t *spa, nvlist_t *nvroot) { uint64_t txg, ndraid = 0; int error; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd, *tvd; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); if ((error = spa_config_parse(spa, &vd, nvroot, NULL, 0, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, NULL, txg, error)); spa->spa_pending_vdev = vd; /* spa_vdev_exit() will clear this */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0) nspares = 0; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) != 0) nl2cache = 0; if (vd->vdev_children == 0 && nspares == 0 && nl2cache == 0) return (spa_vdev_exit(spa, vd, txg, EINVAL)); if (vd->vdev_children != 0 && (error = vdev_create(vd, txg, B_FALSE)) != 0) { return (spa_vdev_exit(spa, vd, txg, error)); } /* * The virtual dRAID spares must be added after vdev tree is created * and the vdev guids are generated. The guid of their associated * dRAID is stored in the config and used when opening the spare. */ if ((error = vdev_draid_spare_create(nvroot, vd, &ndraid, rvd->vdev_children)) == 0) { if (ndraid > 0 && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0) nspares = 0; } else { return (spa_vdev_exit(spa, vd, txg, error)); } /* * We must validate the spares and l2cache devices after checking the * children. Otherwise, vdev_inuse() will blindly overwrite the spare. */ if ((error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, vd, txg, error)); /* * If we are in the middle of a device removal, we can only add * devices which match the existing devices in the pool. * If we are in the middle of a removal, or have some indirect * vdevs, we can not add raidz or dRAID top levels. */ if (spa->spa_vdev_removal != NULL || spa->spa_removing_phys.sr_prev_indirect_vdev != -1) { for (int c = 0; c < vd->vdev_children; c++) { tvd = vd->vdev_child[c]; if (spa->spa_vdev_removal != NULL && tvd->vdev_ashift != spa->spa_max_ashift) { return (spa_vdev_exit(spa, vd, txg, EINVAL)); } /* Fail if top level vdev is raidz or a dRAID */ if (vdev_get_nparity(tvd) != 0) return (spa_vdev_exit(spa, vd, txg, EINVAL)); /* * Need the top level mirror to be * a mirror of leaf vdevs only */ if (tvd->vdev_ops == &vdev_mirror_ops) { for (uint64_t cid = 0; cid < tvd->vdev_children; cid++) { vdev_t *cvd = tvd->vdev_child[cid]; if (!cvd->vdev_ops->vdev_op_leaf) { return (spa_vdev_exit(spa, vd, txg, EINVAL)); } } } } } for (int c = 0; c < vd->vdev_children; c++) { tvd = vd->vdev_child[c]; vdev_remove_child(vd, tvd); tvd->vdev_id = rvd->vdev_children; vdev_add_child(rvd, tvd); vdev_config_dirty(tvd); } if (nspares != 0) { spa_set_aux_vdevs(&spa->spa_spares, spares, nspares, ZPOOL_CONFIG_SPARES); spa_load_spares(spa); spa->spa_spares.sav_sync = B_TRUE; } if (nl2cache != 0) { spa_set_aux_vdevs(&spa->spa_l2cache, l2cache, nl2cache, ZPOOL_CONFIG_L2CACHE); spa_load_l2cache(spa); spa->spa_l2cache.sav_sync = B_TRUE; } /* * We can't increment a feature while holding spa_vdev so we * have to do it in a synctask. */ if (ndraid != 0) { dmu_tx_t *tx; tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); dsl_sync_task_nowait(spa->spa_dsl_pool, spa_draid_feature_incr, (void *)(uintptr_t)ndraid, tx); dmu_tx_commit(tx); } /* * We have to be careful when adding new vdevs to an existing pool. * If other threads start allocating from these vdevs before we * sync the config cache, and we lose power, then upon reboot we may * fail to open the pool because there are DVAs that the config cache * can't translate. Therefore, we first add the vdevs without * initializing metaslabs; sync the config cache (via spa_vdev_exit()); * and then let spa_config_update() initialize the new metaslabs. * * spa_load() checks for added-but-not-initialized vdevs, so that * if we lose power at any point in this sequence, the remaining * steps will be completed the next time we load the pool. */ (void) spa_vdev_exit(spa, vd, txg, 0); mutex_enter(&spa_namespace_lock); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); spa_event_notify(spa, NULL, NULL, ESC_ZFS_VDEV_ADD); mutex_exit(&spa_namespace_lock); return (0); } /* * Attach a device to a mirror. The arguments are the path to any device * in the mirror, and the nvroot for the new device. If the path specifies * a device that is not mirrored, we automatically insert the mirror vdev. * * If 'replacing' is specified, the new device is intended to replace the * existing device; in this case the two devices are made into their own * mirror using the 'replacing' vdev, which is functionally identical to * the mirror vdev (it actually reuses all the same ops) but has a few * extra rules: you can't attach to it after it's been created, and upon * completion of resilvering, the first disk (the one being replaced) * is automatically detached. * * If 'rebuild' is specified, then sequential reconstruction (a.ka. rebuild) * should be performed instead of traditional healing reconstruction. From * an administrators perspective these are both resilver operations. */ int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing, int rebuild) { uint64_t txg, dtl_max_txg; vdev_t *rvd = spa->spa_root_vdev; vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd; vdev_ops_t *pvops; char *oldvdpath, *newvdpath; int newvd_isspare; int error; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); oldvd = spa_lookup_by_guid(spa, guid, B_FALSE); ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) { error = (spa_has_checkpoint(spa)) ? ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT; return (spa_vdev_exit(spa, NULL, txg, error)); } if (rebuild) { if (!spa_feature_is_enabled(spa, SPA_FEATURE_DEVICE_REBUILD)) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); if (dsl_scan_resilvering(spa_get_dsl(spa))) return (spa_vdev_exit(spa, NULL, txg, ZFS_ERR_RESILVER_IN_PROGRESS)); } else { if (vdev_rebuild_active(rvd)) return (spa_vdev_exit(spa, NULL, txg, ZFS_ERR_REBUILD_IN_PROGRESS)); } if (spa->spa_vdev_removal != NULL) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); if (oldvd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!oldvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = oldvd->vdev_parent; if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0, VDEV_ALLOC_ATTACH)) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); if (newrootvd->vdev_children != 1) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); newvd = newrootvd->vdev_child[0]; if (!newvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); if ((error = vdev_create(newrootvd, txg, replacing)) != 0) return (spa_vdev_exit(spa, newrootvd, txg, error)); /* * Spares can't replace logs */ if (oldvd->vdev_top->vdev_islog && newvd->vdev_isspare) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); /* * A dRAID spare can only replace a child of its parent dRAID vdev. */ if (newvd->vdev_ops == &vdev_draid_spare_ops && oldvd->vdev_top != vdev_draid_spare_get_parent(newvd)) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } if (rebuild) { /* * For rebuilds, the top vdev must support reconstruction * using only space maps. This means the only allowable * vdevs types are the root vdev, a mirror, or dRAID. */ tvd = pvd; if (pvd->vdev_top != NULL) tvd = pvd->vdev_top; if (tvd->vdev_ops != &vdev_mirror_ops && tvd->vdev_ops != &vdev_root_ops && tvd->vdev_ops != &vdev_draid_ops) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } } if (!replacing) { /* * For attach, the only allowable parent is a mirror or the root * vdev. */ if (pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_root_ops) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); pvops = &vdev_mirror_ops; } else { /* * Active hot spares can only be replaced by inactive hot * spares. */ if (pvd->vdev_ops == &vdev_spare_ops && oldvd->vdev_isspare && !spa_has_spare(spa, newvd->vdev_guid)) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); /* * If the source is a hot spare, and the parent isn't already a * spare, then we want to create a new hot spare. Otherwise, we * want to create a replacing vdev. The user is not allowed to * attach to a spared vdev child unless the 'isspare' state is * the same (spare replaces spare, non-spare replaces * non-spare). */ if (pvd->vdev_ops == &vdev_replacing_ops && spa_version(spa) < SPA_VERSION_MULTI_REPLACE) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } else if (pvd->vdev_ops == &vdev_spare_ops && newvd->vdev_isspare != oldvd->vdev_isspare) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } if (newvd->vdev_isspare) pvops = &vdev_spare_ops; else pvops = &vdev_replacing_ops; } /* * Make sure the new device is big enough. */ if (newvd->vdev_asize < vdev_get_min_asize(oldvd)) return (spa_vdev_exit(spa, newrootvd, txg, EOVERFLOW)); /* * The new device cannot have a higher alignment requirement * than the top-level vdev. */ if (newvd->vdev_ashift > oldvd->vdev_top->vdev_ashift) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); /* * If this is an in-place replacement, update oldvd's path and devid * to make it distinguishable from newvd, and unopenable from now on. */ if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) { spa_strfree(oldvd->vdev_path); oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5, KM_SLEEP); (void) snprintf(oldvd->vdev_path, strlen(newvd->vdev_path) + 5, "%s/%s", newvd->vdev_path, "old"); if (oldvd->vdev_devid != NULL) { spa_strfree(oldvd->vdev_devid); oldvd->vdev_devid = NULL; } } /* * If the parent is not a mirror, or if we're replacing, insert the new * mirror/replacing/spare vdev above oldvd. */ if (pvd->vdev_ops != pvops) pvd = vdev_add_parent(oldvd, pvops); ASSERT(pvd->vdev_top->vdev_parent == rvd); ASSERT(pvd->vdev_ops == pvops); ASSERT(oldvd->vdev_parent == pvd); /* * Extract the new device from its root and add it to pvd. */ vdev_remove_child(newrootvd, newvd); newvd->vdev_id = pvd->vdev_children; newvd->vdev_crtxg = oldvd->vdev_crtxg; vdev_add_child(pvd, newvd); /* * Reevaluate the parent vdev state. */ vdev_propagate_state(pvd); tvd = newvd->vdev_top; ASSERT(pvd->vdev_top == tvd); ASSERT(tvd->vdev_parent == rvd); vdev_config_dirty(tvd); /* * Set newvd's DTL to [TXG_INITIAL, dtl_max_txg) so that we account * for any dmu_sync-ed blocks. It will propagate upward when * spa_vdev_exit() calls vdev_dtl_reassess(). */ dtl_max_txg = txg + TXG_CONCURRENT_STATES; vdev_dtl_dirty(newvd, DTL_MISSING, TXG_INITIAL, dtl_max_txg - TXG_INITIAL); if (newvd->vdev_isspare) { spa_spare_activate(newvd); spa_event_notify(spa, newvd, NULL, ESC_ZFS_VDEV_SPARE); } oldvdpath = spa_strdup(oldvd->vdev_path); newvdpath = spa_strdup(newvd->vdev_path); newvd_isspare = newvd->vdev_isspare; /* * Mark newvd's DTL dirty in this txg. */ vdev_dirty(tvd, VDD_DTL, newvd, txg); /* * Schedule the resilver or rebuild to restart in the future. We do * this to ensure that dmu_sync-ed blocks have been stitched into the * respective datasets. */ if (rebuild) { newvd->vdev_rebuild_txg = txg; vdev_rebuild(tvd); } else { newvd->vdev_resilver_txg = txg; if (dsl_scan_resilvering(spa_get_dsl(spa)) && spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER)) { vdev_defer_resilver(newvd); } else { dsl_scan_restart_resilver(spa->spa_dsl_pool, dtl_max_txg); } } if (spa->spa_bootfs) spa_event_notify(spa, newvd, NULL, ESC_ZFS_BOOTFS_VDEV_ATTACH); spa_event_notify(spa, newvd, NULL, ESC_ZFS_VDEV_ATTACH); /* * Commit the config */ (void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0); spa_history_log_internal(spa, "vdev attach", NULL, "%s vdev=%s %s vdev=%s", replacing && newvd_isspare ? "spare in" : replacing ? "replace" : "attach", newvdpath, replacing ? "for" : "to", oldvdpath); spa_strfree(oldvdpath); spa_strfree(newvdpath); return (0); } /* * Detach a device from a mirror or replacing vdev. * * If 'replace_done' is specified, only detach if the parent * is a replacing vdev. */ int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) { uint64_t txg; int error; vdev_t *rvd __maybe_unused = spa->spa_root_vdev; vdev_t *vd, *pvd, *cvd, *tvd; boolean_t unspare = B_FALSE; uint64_t unspare_guid = 0; char *vdpath; ASSERT(spa_writeable(spa)); txg = spa_vdev_detach_enter(spa, guid); vd = spa_lookup_by_guid(spa, guid, B_FALSE); /* * Besides being called directly from the userland through the * ioctl interface, spa_vdev_detach() can be potentially called * at the end of spa_vdev_resilver_done(). * * In the regular case, when we have a checkpoint this shouldn't * happen as we never empty the DTLs of a vdev during the scrub * [see comment in dsl_scan_done()]. Thus spa_vdev_resilvering_done() * should never get here when we have a checkpoint. * * That said, even in a case when we checkpoint the pool exactly * as spa_vdev_resilver_done() calls this function everything * should be fine as the resilver will return right away. */ ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) { error = (spa_has_checkpoint(spa)) ? ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT; return (spa_vdev_exit(spa, NULL, txg, error)); } if (vd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = vd->vdev_parent; /* * If the parent/child relationship is not as expected, don't do it. * Consider M(A,R(B,C)) -- that is, a mirror of A with a replacing * vdev that's replacing B with C. The user's intent in replacing * is to go from M(A,B) to M(A,C). If the user decides to cancel * the replace by detaching C, the expected behavior is to end up * M(A,B). But suppose that right after deciding to detach C, * the replacement of B completes. We would have M(A,C), and then * ask to detach C, which would leave us with just A -- not what * the user wanted. To prevent this, we make sure that the * parent/child relationship hasn't changed -- in this example, * that C's parent is still the replacing vdev R. */ if (pvd->vdev_guid != pguid && pguid != 0) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); /* * Only 'replacing' or 'spare' vdevs can be replaced. */ if (replace_done && pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); ASSERT(pvd->vdev_ops != &vdev_spare_ops || spa_version(spa) >= SPA_VERSION_SPARES); /* * Only mirror, replacing, and spare vdevs support detach. */ if (pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); /* * If this device has the only valid copy of some data, * we cannot safely detach it. */ if (vdev_dtl_required(vd)) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); ASSERT(pvd->vdev_children >= 2); /* * If we are detaching the second disk from a replacing vdev, then * check to see if we changed the original vdev's path to have "/old" * at the end in spa_vdev_attach(). If so, undo that change now. */ if (pvd->vdev_ops == &vdev_replacing_ops && vd->vdev_id > 0 && vd->vdev_path != NULL) { size_t len = strlen(vd->vdev_path); for (int c = 0; c < pvd->vdev_children; c++) { cvd = pvd->vdev_child[c]; if (cvd == vd || cvd->vdev_path == NULL) continue; if (strncmp(cvd->vdev_path, vd->vdev_path, len) == 0 && strcmp(cvd->vdev_path + len, "/old") == 0) { spa_strfree(cvd->vdev_path); cvd->vdev_path = spa_strdup(vd->vdev_path); break; } } } /* * If we are detaching the original disk from a normal spare, then it * implies that the spare should become a real disk, and be removed * from the active spare list for the pool. dRAID spares on the * other hand are coupled to the pool and thus should never be removed * from the spares list. */ if (pvd->vdev_ops == &vdev_spare_ops && vd->vdev_id == 0) { vdev_t *last_cvd = pvd->vdev_child[pvd->vdev_children - 1]; if (last_cvd->vdev_isspare && last_cvd->vdev_ops != &vdev_draid_spare_ops) { unspare = B_TRUE; } } /* * Erase the disk labels so the disk can be used for other things. * This must be done after all other error cases are handled, * but before we disembowel vd (so we can still do I/O to it). * But if we can't do it, don't treat the error as fatal -- * it may be that the unwritability of the disk is the reason * it's being detached! */ error = vdev_label_init(vd, 0, VDEV_LABEL_REMOVE); /* * Remove vd from its parent and compact the parent's children. */ vdev_remove_child(pvd, vd); vdev_compact_children(pvd); /* * Remember one of the remaining children so we can get tvd below. */ cvd = pvd->vdev_child[pvd->vdev_children - 1]; /* * If we need to remove the remaining child from the list of hot spares, * do it now, marking the vdev as no longer a spare in the process. * We must do this before vdev_remove_parent(), because that can * change the GUID if it creates a new toplevel GUID. For a similar * reason, we must remove the spare now, in the same txg as the detach; * otherwise someone could attach a new sibling, change the GUID, and * the subsequent attempt to spa_vdev_remove(unspare_guid) would fail. */ if (unspare) { ASSERT(cvd->vdev_isspare); spa_spare_remove(cvd); unspare_guid = cvd->vdev_guid; (void) spa_vdev_remove(spa, unspare_guid, B_TRUE); cvd->vdev_unspare = B_TRUE; } /* * If the parent mirror/replacing vdev only has one child, * the parent is no longer needed. Remove it from the tree. */ if (pvd->vdev_children == 1) { if (pvd->vdev_ops == &vdev_spare_ops) cvd->vdev_unspare = B_FALSE; vdev_remove_parent(cvd); } /* * We don't set tvd until now because the parent we just removed * may have been the previous top-level vdev. */ tvd = cvd->vdev_top; ASSERT(tvd->vdev_parent == rvd); /* * Reevaluate the parent vdev state. */ vdev_propagate_state(cvd); /* * If the 'autoexpand' property is set on the pool then automatically * try to expand the size of the pool. For example if the device we * just detached was smaller than the others, it may be possible to * add metaslabs (i.e. grow the pool). We need to reopen the vdev * first so that we can obtain the updated sizes of the leaf vdevs. */ if (spa->spa_autoexpand) { vdev_reopen(tvd); vdev_expand(tvd, txg); } vdev_config_dirty(tvd); /* * Mark vd's DTL as dirty in this txg. vdev_dtl_sync() will see that * vd->vdev_detached is set and free vd's DTL object in syncing context. * But first make sure we're not on any *other* txg's DTL list, to * prevent vd from being accessed after it's freed. */ vdpath = spa_strdup(vd->vdev_path ? vd->vdev_path : "none"); for (int t = 0; t < TXG_SIZE; t++) (void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t); vd->vdev_detached = B_TRUE; vdev_dirty(tvd, VDD_DTL, vd, txg); spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_REMOVE); spa_notify_waiters(spa); /* hang on to the spa before we release the lock */ spa_open_ref(spa, FTAG); error = spa_vdev_exit(spa, vd, txg, 0); spa_history_log_internal(spa, "detach", NULL, "vdev=%s", vdpath); spa_strfree(vdpath); /* * If this was the removal of the original device in a hot spare vdev, * then we want to go through and remove the device from the hot spare * list of every other pool. */ if (unspare) { spa_t *altspa = NULL; mutex_enter(&spa_namespace_lock); while ((altspa = spa_next(altspa)) != NULL) { if (altspa->spa_state != POOL_STATE_ACTIVE || altspa == spa) continue; spa_open_ref(altspa, FTAG); mutex_exit(&spa_namespace_lock); (void) spa_vdev_remove(altspa, unspare_guid, B_TRUE); mutex_enter(&spa_namespace_lock); spa_close(altspa, FTAG); } mutex_exit(&spa_namespace_lock); /* search the rest of the vdevs for spares to remove */ spa_vdev_resilver_done(spa); } /* all done with the spa; OK to release */ mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); mutex_exit(&spa_namespace_lock); return (error); } static int spa_vdev_initialize_impl(spa_t *spa, uint64_t guid, uint64_t cmd_type, list_t *vd_list) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); /* Look up vdev and ensure it's a leaf. */ vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd == NULL || vd->vdev_detached) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(ENODEV)); } else if (!vd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(vd)) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EINVAL)); } else if (!vdev_writeable(vd)) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EROFS)); } mutex_enter(&vd->vdev_initialize_lock); spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); /* * When we activate an initialize action we check to see * if the vdev_initialize_thread is NULL. We do this instead * of using the vdev_initialize_state since there might be * a previous initialization process which has completed but * the thread is not exited. */ if (cmd_type == POOL_INITIALIZE_START && (vd->vdev_initialize_thread != NULL || vd->vdev_top->vdev_removing)) { mutex_exit(&vd->vdev_initialize_lock); return (SET_ERROR(EBUSY)); } else if (cmd_type == POOL_INITIALIZE_CANCEL && (vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE && vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED)) { mutex_exit(&vd->vdev_initialize_lock); return (SET_ERROR(ESRCH)); } else if (cmd_type == POOL_INITIALIZE_SUSPEND && vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE) { mutex_exit(&vd->vdev_initialize_lock); return (SET_ERROR(ESRCH)); } switch (cmd_type) { case POOL_INITIALIZE_START: vdev_initialize(vd); break; case POOL_INITIALIZE_CANCEL: vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED, vd_list); break; case POOL_INITIALIZE_SUSPEND: vdev_initialize_stop(vd, VDEV_INITIALIZE_SUSPENDED, vd_list); break; default: panic("invalid cmd_type %llu", (unsigned long long)cmd_type); } mutex_exit(&vd->vdev_initialize_lock); return (0); } int spa_vdev_initialize(spa_t *spa, nvlist_t *nv, uint64_t cmd_type, nvlist_t *vdev_errlist) { int total_errors = 0; list_t vd_list; list_create(&vd_list, sizeof (vdev_t), offsetof(vdev_t, vdev_initialize_node)); /* * We hold the namespace lock through the whole function * to prevent any changes to the pool while we're starting or * stopping initialization. The config and state locks are held so that * we can properly assess the vdev state before we commit to * the initializing operation. */ mutex_enter(&spa_namespace_lock); for (nvpair_t *pair = nvlist_next_nvpair(nv, NULL); pair != NULL; pair = nvlist_next_nvpair(nv, pair)) { uint64_t vdev_guid = fnvpair_value_uint64(pair); int error = spa_vdev_initialize_impl(spa, vdev_guid, cmd_type, &vd_list); if (error != 0) { char guid_as_str[MAXNAMELEN]; (void) snprintf(guid_as_str, sizeof (guid_as_str), "%llu", (unsigned long long)vdev_guid); fnvlist_add_int64(vdev_errlist, guid_as_str, error); total_errors++; } } /* Wait for all initialize threads to stop. */ vdev_initialize_stop_wait(spa, &vd_list); /* Sync out the initializing state */ txg_wait_synced(spa->spa_dsl_pool, 0); mutex_exit(&spa_namespace_lock); list_destroy(&vd_list); return (total_errors); } static int spa_vdev_trim_impl(spa_t *spa, uint64_t guid, uint64_t cmd_type, uint64_t rate, boolean_t partial, boolean_t secure, list_t *vd_list) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); /* Look up vdev and ensure it's a leaf. */ vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd == NULL || vd->vdev_detached) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(ENODEV)); } else if (!vd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(vd)) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EINVAL)); } else if (!vdev_writeable(vd)) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EROFS)); } else if (!vd->vdev_has_trim) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EOPNOTSUPP)); } else if (secure && !vd->vdev_has_securetrim) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (SET_ERROR(EOPNOTSUPP)); } mutex_enter(&vd->vdev_trim_lock); spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); /* * When we activate a TRIM action we check to see if the * vdev_trim_thread is NULL. We do this instead of using the * vdev_trim_state since there might be a previous TRIM process * which has completed but the thread is not exited. */ if (cmd_type == POOL_TRIM_START && (vd->vdev_trim_thread != NULL || vd->vdev_top->vdev_removing)) { mutex_exit(&vd->vdev_trim_lock); return (SET_ERROR(EBUSY)); } else if (cmd_type == POOL_TRIM_CANCEL && (vd->vdev_trim_state != VDEV_TRIM_ACTIVE && vd->vdev_trim_state != VDEV_TRIM_SUSPENDED)) { mutex_exit(&vd->vdev_trim_lock); return (SET_ERROR(ESRCH)); } else if (cmd_type == POOL_TRIM_SUSPEND && vd->vdev_trim_state != VDEV_TRIM_ACTIVE) { mutex_exit(&vd->vdev_trim_lock); return (SET_ERROR(ESRCH)); } switch (cmd_type) { case POOL_TRIM_START: vdev_trim(vd, rate, partial, secure); break; case POOL_TRIM_CANCEL: vdev_trim_stop(vd, VDEV_TRIM_CANCELED, vd_list); break; case POOL_TRIM_SUSPEND: vdev_trim_stop(vd, VDEV_TRIM_SUSPENDED, vd_list); break; default: panic("invalid cmd_type %llu", (unsigned long long)cmd_type); } mutex_exit(&vd->vdev_trim_lock); return (0); } /* * Initiates a manual TRIM for the requested vdevs. This kicks off individual * TRIM threads for each child vdev. These threads pass over all of the free * space in the vdev's metaslabs and issues TRIM commands for that space. */ int spa_vdev_trim(spa_t *spa, nvlist_t *nv, uint64_t cmd_type, uint64_t rate, boolean_t partial, boolean_t secure, nvlist_t *vdev_errlist) { int total_errors = 0; list_t vd_list; list_create(&vd_list, sizeof (vdev_t), offsetof(vdev_t, vdev_trim_node)); /* * We hold the namespace lock through the whole function * to prevent any changes to the pool while we're starting or * stopping TRIM. The config and state locks are held so that * we can properly assess the vdev state before we commit to * the TRIM operation. */ mutex_enter(&spa_namespace_lock); for (nvpair_t *pair = nvlist_next_nvpair(nv, NULL); pair != NULL; pair = nvlist_next_nvpair(nv, pair)) { uint64_t vdev_guid = fnvpair_value_uint64(pair); int error = spa_vdev_trim_impl(spa, vdev_guid, cmd_type, rate, partial, secure, &vd_list); if (error != 0) { char guid_as_str[MAXNAMELEN]; (void) snprintf(guid_as_str, sizeof (guid_as_str), "%llu", (unsigned long long)vdev_guid); fnvlist_add_int64(vdev_errlist, guid_as_str, error); total_errors++; } } /* Wait for all TRIM threads to stop. */ vdev_trim_stop_wait(spa, &vd_list); /* Sync out the TRIM state */ txg_wait_synced(spa->spa_dsl_pool, 0); mutex_exit(&spa_namespace_lock); list_destroy(&vd_list); return (total_errors); } /* * Split a set of devices from their mirrors, and create a new pool from them. */ int spa_vdev_split_mirror(spa_t *spa, const char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp) { int error = 0; uint64_t txg, *glist; spa_t *newspa; uint_t c, children, lastlog; nvlist_t **child, *nvl, *tmp; dmu_tx_t *tx; char *altroot = NULL; vdev_t *rvd, **vml = NULL; /* vdev modify list */ boolean_t activate_slog; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) { error = (spa_has_checkpoint(spa)) ? ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT; return (spa_vdev_exit(spa, NULL, txg, error)); } /* clear the log and flush everything up to now */ activate_slog = spa_passivate_log(spa); (void) spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); error = spa_reset_logs(spa); txg = spa_vdev_config_enter(spa); if (activate_slog) spa_activate_log(spa); if (error != 0) return (spa_vdev_exit(spa, NULL, txg, error)); /* check new spa name before going any further */ if (spa_lookup(newname) != NULL) return (spa_vdev_exit(spa, NULL, txg, EEXIST)); /* * scan through all the children to ensure they're all mirrors */ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* first, check to ensure we've got the right child count */ rvd = spa->spa_root_vdev; lastlog = 0; for (c = 0; c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; /* don't count the holes & logs as children */ if (vd->vdev_islog || (vd->vdev_ops != &vdev_indirect_ops && !vdev_is_concrete(vd))) { if (lastlog == 0) lastlog = c; continue; } lastlog = 0; } if (children != (lastlog != 0 ? lastlog : rvd->vdev_children)) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* next, ensure no spare or cache devices are part of the split */ if (nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_SPARES, &tmp) == 0 || nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); vml = kmem_zalloc(children * sizeof (vdev_t *), KM_SLEEP); glist = kmem_zalloc(children * sizeof (uint64_t), KM_SLEEP); /* then, loop over each vdev and validate it */ for (c = 0; c < children; c++) { uint64_t is_hole = 0; (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE, &is_hole); if (is_hole != 0) { if (spa->spa_root_vdev->vdev_child[c]->vdev_ishole || spa->spa_root_vdev->vdev_child[c]->vdev_islog) { continue; } else { error = SET_ERROR(EINVAL); break; } } /* deal with indirect vdevs */ if (spa->spa_root_vdev->vdev_child[c]->vdev_ops == &vdev_indirect_ops) continue; /* which disk is going to be split? */ if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_GUID, &glist[c]) != 0) { error = SET_ERROR(EINVAL); break; } /* look it up in the spa */ vml[c] = spa_lookup_by_guid(spa, glist[c], B_FALSE); if (vml[c] == NULL) { error = SET_ERROR(ENODEV); break; } /* make sure there's nothing stopping the split */ if (vml[c]->vdev_parent->vdev_ops != &vdev_mirror_ops || vml[c]->vdev_islog || !vdev_is_concrete(vml[c]) || vml[c]->vdev_isspare || vml[c]->vdev_isl2cache || !vdev_writeable(vml[c]) || vml[c]->vdev_children != 0 || vml[c]->vdev_state != VDEV_STATE_HEALTHY || c != spa->spa_root_vdev->vdev_child[c]->vdev_id) { error = SET_ERROR(EINVAL); break; } if (vdev_dtl_required(vml[c]) || vdev_resilver_needed(vml[c], NULL, NULL)) { error = SET_ERROR(EBUSY); break; } /* we need certain info from the top level */ fnvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_ARRAY, vml[c]->vdev_top->vdev_ms_array); fnvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_SHIFT, vml[c]->vdev_top->vdev_ms_shift); fnvlist_add_uint64(child[c], ZPOOL_CONFIG_ASIZE, vml[c]->vdev_top->vdev_asize); fnvlist_add_uint64(child[c], ZPOOL_CONFIG_ASHIFT, vml[c]->vdev_top->vdev_ashift); /* transfer per-vdev ZAPs */ ASSERT3U(vml[c]->vdev_leaf_zap, !=, 0); VERIFY0(nvlist_add_uint64(child[c], ZPOOL_CONFIG_VDEV_LEAF_ZAP, vml[c]->vdev_leaf_zap)); ASSERT3U(vml[c]->vdev_top->vdev_top_zap, !=, 0); VERIFY0(nvlist_add_uint64(child[c], ZPOOL_CONFIG_VDEV_TOP_ZAP, vml[c]->vdev_parent->vdev_top_zap)); } if (error != 0) { kmem_free(vml, children * sizeof (vdev_t *)); kmem_free(glist, children * sizeof (uint64_t)); return (spa_vdev_exit(spa, NULL, txg, error)); } /* stop writers from using the disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_TRUE; } vdev_reopen(spa->spa_root_vdev); /* * Temporarily record the splitting vdevs in the spa config. This * will disappear once the config is regenerated. */ nvl = fnvlist_alloc(); fnvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, glist, children); kmem_free(glist, children * sizeof (uint64_t)); mutex_enter(&spa->spa_props_lock); fnvlist_add_nvlist(spa->spa_config, ZPOOL_CONFIG_SPLIT, nvl); mutex_exit(&spa->spa_props_lock); spa->spa_config_splitting = nvl; vdev_config_dirty(spa->spa_root_vdev); /* configure and create the new pool */ fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, newname); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, exp ? POOL_STATE_EXPORTED : POOL_STATE_ACTIVE); fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, spa->spa_config_txg); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_generate_guid(NULL)); VERIFY0(nvlist_add_boolean(config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS)); (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); /* add the new pool to the namespace */ newspa = spa_add(newname, config, altroot); newspa->spa_avz_action = AVZ_ACTION_REBUILD; newspa->spa_config_txg = spa->spa_config_txg; spa_set_log_state(newspa, SPA_LOG_CLEAR); /* release the spa config lock, retaining the namespace lock */ spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 1); spa_activate(newspa, spa_mode_global); spa_async_suspend(newspa); /* * Temporarily stop the initializing and TRIM activity. We set the * state to ACTIVE so that we know to resume initializing or TRIM * once the split has completed. */ list_t vd_initialize_list; list_create(&vd_initialize_list, sizeof (vdev_t), offsetof(vdev_t, vdev_initialize_node)); list_t vd_trim_list; list_create(&vd_trim_list, sizeof (vdev_t), offsetof(vdev_t, vdev_trim_node)); for (c = 0; c < children; c++) { if (vml[c] != NULL && vml[c]->vdev_ops != &vdev_indirect_ops) { mutex_enter(&vml[c]->vdev_initialize_lock); vdev_initialize_stop(vml[c], VDEV_INITIALIZE_ACTIVE, &vd_initialize_list); mutex_exit(&vml[c]->vdev_initialize_lock); mutex_enter(&vml[c]->vdev_trim_lock); vdev_trim_stop(vml[c], VDEV_TRIM_ACTIVE, &vd_trim_list); mutex_exit(&vml[c]->vdev_trim_lock); } } vdev_initialize_stop_wait(spa, &vd_initialize_list); vdev_trim_stop_wait(spa, &vd_trim_list); list_destroy(&vd_initialize_list); list_destroy(&vd_trim_list); newspa->spa_config_source = SPA_CONFIG_SRC_SPLIT; newspa->spa_is_splitting = B_TRUE; /* create the new pool from the disks of the original pool */ error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE); if (error) goto out; /* if that worked, generate a real config for the new pool */ if (newspa->spa_root_vdev != NULL) { newspa->spa_config_splitting = fnvlist_alloc(); fnvlist_add_uint64(newspa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa)); spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL, B_TRUE)); } /* set the props */ if (props != NULL) { spa_configfile_set(newspa, props, B_FALSE); error = spa_prop_set(newspa, props); if (error) goto out; } /* flush everything */ txg = spa_vdev_config_enter(newspa); vdev_config_dirty(newspa->spa_root_vdev); (void) spa_vdev_config_exit(newspa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 2); spa_async_resume(newspa); /* finally, update the original pool's config */ txg = spa_vdev_config_enter(spa); tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) dmu_tx_abort(tx); for (c = 0; c < children; c++) { if (vml[c] != NULL && vml[c]->vdev_ops != &vdev_indirect_ops) { vdev_t *tvd = vml[c]->vdev_top; /* * Need to be sure the detachable VDEV is not * on any *other* txg's DTL list to prevent it * from being accessed after it's freed. */ for (int t = 0; t < TXG_SIZE; t++) { (void) txg_list_remove_this( &tvd->vdev_dtl_list, vml[c], t); } vdev_split(vml[c]); if (error == 0) spa_history_log_internal(spa, "detach", tx, "vdev=%s", vml[c]->vdev_path); vdev_free(vml[c]); } } spa->spa_avz_action = AVZ_ACTION_REBUILD; vdev_config_dirty(spa->spa_root_vdev); spa->spa_config_splitting = NULL; nvlist_free(nvl); if (error == 0) dmu_tx_commit(tx); (void) spa_vdev_exit(spa, NULL, txg, 0); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 3); /* split is complete; log a history record */ spa_history_log_internal(newspa, "split", NULL, "from pool %s", spa_name(spa)); newspa->spa_is_splitting = B_FALSE; kmem_free(vml, children * sizeof (vdev_t *)); /* if we're not going to mount the filesystems in userland, export */ if (exp) error = spa_export_common(newname, POOL_STATE_EXPORTED, NULL, B_FALSE, B_FALSE); return (error); out: spa_unload(newspa); spa_deactivate(newspa); spa_remove(newspa); txg = spa_vdev_config_enter(spa); /* re-online all offlined disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_FALSE; } /* restart initializing or trimming disks as necessary */ spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART); spa_async_request(spa, SPA_ASYNC_TRIM_RESTART); spa_async_request(spa, SPA_ASYNC_AUTOTRIM_RESTART); vdev_reopen(spa->spa_root_vdev); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; (void) spa_vdev_exit(spa, NULL, txg, error); kmem_free(vml, children * sizeof (vdev_t *)); return (error); } /* * Find any device that's done replacing, or a vdev marked 'unspare' that's * currently spared, so we can detach it. */ static vdev_t * spa_vdev_resilver_done_hunt(vdev_t *vd) { vdev_t *newvd, *oldvd; for (int c = 0; c < vd->vdev_children; c++) { oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]); if (oldvd != NULL) return (oldvd); } /* * Check for a completed replacement. We always consider the first * vdev in the list to be the oldest vdev, and the last one to be * the newest (see spa_vdev_attach() for how that works). In * the case where the newest vdev is faulted, we will not automatically * remove it after a resilver completes. This is OK as it will require * user intervention to determine which disk the admin wishes to keep. */ if (vd->vdev_ops == &vdev_replacing_ops) { ASSERT(vd->vdev_children > 1); newvd = vd->vdev_child[vd->vdev_children - 1]; oldvd = vd->vdev_child[0]; if (vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); } /* * Check for a completed resilver with the 'unspare' flag set. * Also potentially update faulted state. */ if (vd->vdev_ops == &vdev_spare_ops) { vdev_t *first = vd->vdev_child[0]; vdev_t *last = vd->vdev_child[vd->vdev_children - 1]; if (last->vdev_unspare) { oldvd = first; newvd = last; } else if (first->vdev_unspare) { oldvd = last; newvd = first; } else { oldvd = NULL; } if (oldvd != NULL && vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); vdev_propagate_state(vd); /* * If there are more than two spares attached to a disk, * and those spares are not required, then we want to * attempt to free them up now so that they can be used * by other pools. Once we're back down to a single * disk+spare, we stop removing them. */ if (vd->vdev_children > 2) { newvd = vd->vdev_child[1]; if (newvd->vdev_isspare && last->vdev_isspare && vdev_dtl_empty(last, DTL_MISSING) && vdev_dtl_empty(last, DTL_OUTAGE) && !vdev_dtl_required(newvd)) return (newvd); } } return (NULL); } static void spa_vdev_resilver_done(spa_t *spa) { vdev_t *vd, *pvd, *ppvd; uint64_t guid, sguid, pguid, ppguid; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); while ((vd = spa_vdev_resilver_done_hunt(spa->spa_root_vdev)) != NULL) { pvd = vd->vdev_parent; ppvd = pvd->vdev_parent; guid = vd->vdev_guid; pguid = pvd->vdev_guid; ppguid = ppvd->vdev_guid; sguid = 0; /* * If we have just finished replacing a hot spared device, then * we need to detach the parent's first child (the original hot * spare) as well. */ if (ppvd->vdev_ops == &vdev_spare_ops && pvd->vdev_id == 0 && ppvd->vdev_children == 2) { ASSERT(pvd->vdev_ops == &vdev_replacing_ops); sguid = ppvd->vdev_child[1]->vdev_guid; } ASSERT(vd->vdev_resilver_txg == 0 || !vdev_dtl_required(vd)); spa_config_exit(spa, SCL_ALL, FTAG); if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0) return; if (sguid && spa_vdev_detach(spa, sguid, ppguid, B_TRUE) != 0) return; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); } spa_config_exit(spa, SCL_ALL, FTAG); /* * If a detach was not performed above replace waiters will not have * been notified. In which case we must do so now. */ spa_notify_waiters(spa); } /* * Update the stored path or FRU for this vdev. */ static int spa_vdev_set_common(spa_t *spa, uint64_t guid, const char *value, boolean_t ispath) { vdev_t *vd; boolean_t sync = B_FALSE; ASSERT(spa_writeable(spa)); spa_vdev_state_enter(spa, SCL_ALL); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENOENT)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); if (ispath) { if (strcmp(value, vd->vdev_path) != 0) { spa_strfree(vd->vdev_path); vd->vdev_path = spa_strdup(value); sync = B_TRUE; } } else { if (vd->vdev_fru == NULL) { vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } else if (strcmp(value, vd->vdev_fru) != 0) { spa_strfree(vd->vdev_fru); vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } } return (spa_vdev_state_exit(spa, sync ? vd : NULL, 0)); } int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath) { return (spa_vdev_set_common(spa, guid, newpath, B_TRUE)); } int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru) { return (spa_vdev_set_common(spa, guid, newfru, B_FALSE)); } /* * ========================================================================== * SPA Scanning * ========================================================================== */ int spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t cmd) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (dsl_scan_resilvering(spa->spa_dsl_pool)) return (SET_ERROR(EBUSY)); return (dsl_scrub_set_pause_resume(spa->spa_dsl_pool, cmd)); } int spa_scan_stop(spa_t *spa) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (dsl_scan_resilvering(spa->spa_dsl_pool)) return (SET_ERROR(EBUSY)); return (dsl_scan_cancel(spa->spa_dsl_pool)); } int spa_scan(spa_t *spa, pool_scan_func_t func) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (func >= POOL_SCAN_FUNCS || func == POOL_SCAN_NONE) return (SET_ERROR(ENOTSUP)); if (func == POOL_SCAN_RESILVER && !spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER)) return (SET_ERROR(ENOTSUP)); /* * If a resilver was requested, but there is no DTL on a * writeable leaf device, we have nothing to do. */ if (func == POOL_SCAN_RESILVER && !vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) { spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); return (0); } return (dsl_scan(spa->spa_dsl_pool, func)); } /* * ========================================================================== * SPA async task processing * ========================================================================== */ static void spa_async_remove(spa_t *spa, vdev_t *vd) { if (vd->vdev_remove_wanted) { vd->vdev_remove_wanted = B_FALSE; vd->vdev_delayed_close = B_FALSE; vdev_set_state(vd, B_FALSE, VDEV_STATE_REMOVED, VDEV_AUX_NONE); /* * We want to clear the stats, but we don't want to do a full * vdev_clear() as that will cause us to throw away * degraded/faulted state as well as attempt to reopen the * device, all of which is a waste. */ vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; vdev_state_dirty(vd->vdev_top); /* Tell userspace that the vdev is gone. */ zfs_post_remove(spa, vd); } for (int c = 0; c < vd->vdev_children; c++) spa_async_remove(spa, vd->vdev_child[c]); } static void spa_async_probe(spa_t *spa, vdev_t *vd) { if (vd->vdev_probe_wanted) { vd->vdev_probe_wanted = B_FALSE; vdev_reopen(vd); /* vdev_open() does the actual probe */ } for (int c = 0; c < vd->vdev_children; c++) spa_async_probe(spa, vd->vdev_child[c]); } static void spa_async_autoexpand(spa_t *spa, vdev_t *vd) { if (!spa->spa_autoexpand) return; for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; spa_async_autoexpand(spa, cvd); } if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL) return; spa_event_notify(vd->vdev_spa, vd, NULL, ESC_ZFS_VDEV_AUTOEXPAND); } static __attribute__((noreturn)) void spa_async_thread(void *arg) { spa_t *spa = (spa_t *)arg; dsl_pool_t *dp = spa->spa_dsl_pool; int tasks; ASSERT(spa->spa_sync_on); mutex_enter(&spa->spa_async_lock); tasks = spa->spa_async_tasks; spa->spa_async_tasks = 0; mutex_exit(&spa->spa_async_lock); /* * See if the config needs to be updated. */ if (tasks & SPA_ASYNC_CONFIG_UPDATE) { uint64_t old_space, new_space; mutex_enter(&spa_namespace_lock); old_space = metaslab_class_get_space(spa_normal_class(spa)); old_space += metaslab_class_get_space(spa_special_class(spa)); old_space += metaslab_class_get_space(spa_dedup_class(spa)); old_space += metaslab_class_get_space( spa_embedded_log_class(spa)); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); new_space = metaslab_class_get_space(spa_normal_class(spa)); new_space += metaslab_class_get_space(spa_special_class(spa)); new_space += metaslab_class_get_space(spa_dedup_class(spa)); new_space += metaslab_class_get_space( spa_embedded_log_class(spa)); mutex_exit(&spa_namespace_lock); /* * If the pool grew as a result of the config update, * then log an internal history event. */ if (new_space != old_space) { spa_history_log_internal(spa, "vdev online", NULL, "pool '%s' size: %llu(+%llu)", spa_name(spa), (u_longlong_t)new_space, (u_longlong_t)(new_space - old_space)); } } /* * See if any devices need to be marked REMOVED. */ if (tasks & SPA_ASYNC_REMOVE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_remove(spa, spa->spa_root_vdev); for (int i = 0; i < spa->spa_l2cache.sav_count; i++) spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]); for (int i = 0; i < spa->spa_spares.sav_count; i++) spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]); (void) spa_vdev_state_exit(spa, NULL, 0); } if ((tasks & SPA_ASYNC_AUTOEXPAND) && !spa_suspended(spa)) { spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_async_autoexpand(spa, spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); } /* * See if any devices need to be probed. */ if (tasks & SPA_ASYNC_PROBE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_probe(spa, spa->spa_root_vdev); (void) spa_vdev_state_exit(spa, NULL, 0); } /* * If any devices are done replacing, detach them. */ if (tasks & SPA_ASYNC_RESILVER_DONE || tasks & SPA_ASYNC_REBUILD_DONE) { spa_vdev_resilver_done(spa); } /* * Kick off a resilver. */ if (tasks & SPA_ASYNC_RESILVER && !vdev_rebuild_active(spa->spa_root_vdev) && (!dsl_scan_resilvering(dp) || !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER))) dsl_scan_restart_resilver(dp, 0); if (tasks & SPA_ASYNC_INITIALIZE_RESTART) { mutex_enter(&spa_namespace_lock); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); vdev_initialize_restart(spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); mutex_exit(&spa_namespace_lock); } if (tasks & SPA_ASYNC_TRIM_RESTART) { mutex_enter(&spa_namespace_lock); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); vdev_trim_restart(spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); mutex_exit(&spa_namespace_lock); } if (tasks & SPA_ASYNC_AUTOTRIM_RESTART) { mutex_enter(&spa_namespace_lock); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); vdev_autotrim_restart(spa); spa_config_exit(spa, SCL_CONFIG, FTAG); mutex_exit(&spa_namespace_lock); } /* * Kick off L2 cache whole device TRIM. */ if (tasks & SPA_ASYNC_L2CACHE_TRIM) { mutex_enter(&spa_namespace_lock); spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); vdev_trim_l2arc(spa); spa_config_exit(spa, SCL_CONFIG, FTAG); mutex_exit(&spa_namespace_lock); } /* * Kick off L2 cache rebuilding. */ if (tasks & SPA_ASYNC_L2CACHE_REBUILD) { mutex_enter(&spa_namespace_lock); spa_config_enter(spa, SCL_L2ARC, FTAG, RW_READER); l2arc_spa_rebuild_start(spa); spa_config_exit(spa, SCL_L2ARC, FTAG); mutex_exit(&spa_namespace_lock); } /* * Let the world know that we're done. */ mutex_enter(&spa->spa_async_lock); spa->spa_async_thread = NULL; cv_broadcast(&spa->spa_async_cv); mutex_exit(&spa->spa_async_lock); thread_exit(); } void spa_async_suspend(spa_t *spa) { mutex_enter(&spa->spa_async_lock); spa->spa_async_suspended++; while (spa->spa_async_thread != NULL) cv_wait(&spa->spa_async_cv, &spa->spa_async_lock); mutex_exit(&spa->spa_async_lock); spa_vdev_remove_suspend(spa); zthr_t *condense_thread = spa->spa_condense_zthr; if (condense_thread != NULL) zthr_cancel(condense_thread); zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr; if (discard_thread != NULL) zthr_cancel(discard_thread); zthr_t *ll_delete_thread = spa->spa_livelist_delete_zthr; if (ll_delete_thread != NULL) zthr_cancel(ll_delete_thread); zthr_t *ll_condense_thread = spa->spa_livelist_condense_zthr; if (ll_condense_thread != NULL) zthr_cancel(ll_condense_thread); } void spa_async_resume(spa_t *spa) { mutex_enter(&spa->spa_async_lock); ASSERT(spa->spa_async_suspended != 0); spa->spa_async_suspended--; mutex_exit(&spa->spa_async_lock); spa_restart_removal(spa); zthr_t *condense_thread = spa->spa_condense_zthr; if (condense_thread != NULL) zthr_resume(condense_thread); zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr; if (discard_thread != NULL) zthr_resume(discard_thread); zthr_t *ll_delete_thread = spa->spa_livelist_delete_zthr; if (ll_delete_thread != NULL) zthr_resume(ll_delete_thread); zthr_t *ll_condense_thread = spa->spa_livelist_condense_zthr; if (ll_condense_thread != NULL) zthr_resume(ll_condense_thread); } static boolean_t spa_async_tasks_pending(spa_t *spa) { uint_t non_config_tasks; uint_t config_task; boolean_t config_task_suspended; non_config_tasks = spa->spa_async_tasks & ~SPA_ASYNC_CONFIG_UPDATE; config_task = spa->spa_async_tasks & SPA_ASYNC_CONFIG_UPDATE; if (spa->spa_ccw_fail_time == 0) { config_task_suspended = B_FALSE; } else { config_task_suspended = (gethrtime() - spa->spa_ccw_fail_time) < ((hrtime_t)zfs_ccw_retry_interval * NANOSEC); } return (non_config_tasks || (config_task && !config_task_suspended)); } static void spa_async_dispatch(spa_t *spa) { mutex_enter(&spa->spa_async_lock); if (spa_async_tasks_pending(spa) && !spa->spa_async_suspended && spa->spa_async_thread == NULL) spa->spa_async_thread = thread_create(NULL, 0, spa_async_thread, spa, 0, &p0, TS_RUN, maxclsyspri); mutex_exit(&spa->spa_async_lock); } void spa_async_request(spa_t *spa, int task) { zfs_dbgmsg("spa=%s async request task=%u", spa->spa_name, task); mutex_enter(&spa->spa_async_lock); spa->spa_async_tasks |= task; mutex_exit(&spa->spa_async_lock); } int spa_async_tasks(spa_t *spa) { return (spa->spa_async_tasks); } /* * ========================================================================== * SPA syncing routines * ========================================================================== */ static int bpobj_enqueue_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx) { bpobj_t *bpo = arg; bpobj_enqueue(bpo, bp, bp_freed, tx); return (0); } int bpobj_enqueue_alloc_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { return (bpobj_enqueue_cb(arg, bp, B_FALSE, tx)); } int bpobj_enqueue_free_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { return (bpobj_enqueue_cb(arg, bp, B_TRUE, tx)); } static int spa_free_sync_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { zio_t *pio = arg; zio_nowait(zio_free_sync(pio, pio->io_spa, dmu_tx_get_txg(tx), bp, pio->io_flags)); return (0); } static int bpobj_spa_free_sync_cb(void *arg, const blkptr_t *bp, boolean_t bp_freed, dmu_tx_t *tx) { ASSERT(!bp_freed); return (spa_free_sync_cb(arg, bp, tx)); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing frees. */ static void spa_sync_frees(spa_t *spa, bplist_t *bpl, dmu_tx_t *tx) { zio_t *zio = zio_root(spa, NULL, NULL, 0); bplist_iterate(bpl, spa_free_sync_cb, zio, tx); VERIFY(zio_wait(zio) == 0); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing deferred frees. */ static void spa_sync_deferred_frees(spa_t *spa, dmu_tx_t *tx) { if (spa_sync_pass(spa) != 1) return; /* * Note: * If the log space map feature is active, we stop deferring * frees to the next TXG and therefore running this function * would be considered a no-op as spa_deferred_bpobj should * not have any entries. * * That said we run this function anyway (instead of returning * immediately) for the edge-case scenario where we just * activated the log space map feature in this TXG but we have * deferred frees from the previous TXG. */ zio_t *zio = zio_root(spa, NULL, NULL, 0); VERIFY3U(bpobj_iterate(&spa->spa_deferred_bpobj, bpobj_spa_free_sync_cb, zio, tx), ==, 0); VERIFY0(zio_wait(zio)); } static void spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx) { char *packed = NULL; size_t bufsize; size_t nvsize = 0; dmu_buf_t *db; VERIFY(nvlist_size(nv, &nvsize, NV_ENCODE_XDR) == 0); /* * Write full (SPA_CONFIG_BLOCKSIZE) blocks of configuration * information. This avoids the dmu_buf_will_dirty() path and * saves us a pre-read to get data we don't actually care about. */ bufsize = P2ROUNDUP((uint64_t)nvsize, SPA_CONFIG_BLOCKSIZE); packed = vmem_alloc(bufsize, KM_SLEEP); VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR, KM_SLEEP) == 0); memset(packed + nvsize, 0, bufsize - nvsize); dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx); vmem_free(packed, bufsize); VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db)); dmu_buf_will_dirty(db, tx); *(uint64_t *)db->db_data = nvsize; dmu_buf_rele(db, FTAG); } static void spa_sync_aux_dev(spa_t *spa, spa_aux_vdev_t *sav, dmu_tx_t *tx, const char *config, const char *entry) { nvlist_t *nvroot; nvlist_t **list; int i; if (!sav->sav_sync) return; /* * Update the MOS nvlist describing the list of available devices. * spa_validate_aux() will have already made sure this nvlist is * valid and the vdevs are labeled appropriately. */ if (sav->sav_object == 0) { sav->sav_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, 1 << 14, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); VERIFY(zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, entry, sizeof (uint64_t), 1, &sav->sav_object, tx) == 0); } nvroot = fnvlist_alloc(); if (sav->sav_count == 0) { fnvlist_add_nvlist_array(nvroot, config, (const nvlist_t * const *)NULL, 0); } else { list = kmem_alloc(sav->sav_count*sizeof (void *), KM_SLEEP); for (i = 0; i < sav->sav_count; i++) list[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_FALSE, VDEV_CONFIG_L2CACHE); fnvlist_add_nvlist_array(nvroot, config, (const nvlist_t * const *)list, sav->sav_count); for (i = 0; i < sav->sav_count; i++) nvlist_free(list[i]); kmem_free(list, sav->sav_count * sizeof (void *)); } spa_sync_nvlist(spa, sav->sav_object, nvroot, tx); nvlist_free(nvroot); sav->sav_sync = B_FALSE; } /* * Rebuild spa's all-vdev ZAP from the vdev ZAPs indicated in each vdev_t. * The all-vdev ZAP must be empty. */ static void spa_avz_build(vdev_t *vd, uint64_t avz, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; if (vd->vdev_top_zap != 0) { VERIFY0(zap_add_int(spa->spa_meta_objset, avz, vd->vdev_top_zap, tx)); } if (vd->vdev_leaf_zap != 0) { VERIFY0(zap_add_int(spa->spa_meta_objset, avz, vd->vdev_leaf_zap, tx)); } for (uint64_t i = 0; i < vd->vdev_children; i++) { spa_avz_build(vd->vdev_child[i], avz, tx); } } static void spa_sync_config_object(spa_t *spa, dmu_tx_t *tx) { nvlist_t *config; /* * If the pool is being imported from a pre-per-vdev-ZAP version of ZFS, * its config may not be dirty but we still need to build per-vdev ZAPs. * Similarly, if the pool is being assembled (e.g. after a split), we * need to rebuild the AVZ although the config may not be dirty. */ if (list_is_empty(&spa->spa_config_dirty_list) && spa->spa_avz_action == AVZ_ACTION_NONE) return; spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); ASSERT(spa->spa_avz_action == AVZ_ACTION_NONE || spa->spa_avz_action == AVZ_ACTION_INITIALIZE || spa->spa_all_vdev_zaps != 0); if (spa->spa_avz_action == AVZ_ACTION_REBUILD) { /* Make and build the new AVZ */ uint64_t new_avz = zap_create(spa->spa_meta_objset, DMU_OTN_ZAP_METADATA, DMU_OT_NONE, 0, tx); spa_avz_build(spa->spa_root_vdev, new_avz, tx); /* Diff old AVZ with new one */ zap_cursor_t zc; zap_attribute_t za; for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_all_vdev_zaps); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { uint64_t vdzap = za.za_first_integer; if (zap_lookup_int(spa->spa_meta_objset, new_avz, vdzap) == ENOENT) { /* * ZAP is listed in old AVZ but not in new one; * destroy it */ VERIFY0(zap_destroy(spa->spa_meta_objset, vdzap, tx)); } } zap_cursor_fini(&zc); /* Destroy the old AVZ */ VERIFY0(zap_destroy(spa->spa_meta_objset, spa->spa_all_vdev_zaps, tx)); /* Replace the old AVZ in the dir obj with the new one */ VERIFY0(zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_VDEV_ZAP_MAP, sizeof (new_avz), 1, &new_avz, tx)); spa->spa_all_vdev_zaps = new_avz; } else if (spa->spa_avz_action == AVZ_ACTION_DESTROY) { zap_cursor_t zc; zap_attribute_t za; /* Walk through the AVZ and destroy all listed ZAPs */ for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_all_vdev_zaps); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { uint64_t zap = za.za_first_integer; VERIFY0(zap_destroy(spa->spa_meta_objset, zap, tx)); } zap_cursor_fini(&zc); /* Destroy and unlink the AVZ itself */ VERIFY0(zap_destroy(spa->spa_meta_objset, spa->spa_all_vdev_zaps, tx)); VERIFY0(zap_remove(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_VDEV_ZAP_MAP, tx)); spa->spa_all_vdev_zaps = 0; } if (spa->spa_all_vdev_zaps == 0) { spa->spa_all_vdev_zaps = zap_create_link(spa->spa_meta_objset, DMU_OTN_ZAP_METADATA, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_VDEV_ZAP_MAP, tx); } spa->spa_avz_action = AVZ_ACTION_NONE; /* Create ZAPs for vdevs that don't have them. */ vdev_construct_zaps(spa->spa_root_vdev, tx); config = spa_config_generate(spa, spa->spa_root_vdev, dmu_tx_get_txg(tx), B_FALSE); /* * If we're upgrading the spa version then make sure that * the config object gets updated with the correct version. */ if (spa->spa_ubsync.ub_version < spa->spa_uberblock.ub_version) fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa->spa_uberblock.ub_version); spa_config_exit(spa, SCL_STATE, FTAG); nvlist_free(spa->spa_config_syncing); spa->spa_config_syncing = config; spa_sync_nvlist(spa, spa->spa_config_object, config, tx); } static void spa_sync_version(void *arg, dmu_tx_t *tx) { uint64_t *versionp = arg; uint64_t version = *versionp; spa_t *spa = dmu_tx_pool(tx)->dp_spa; /* * Setting the version is special cased when first creating the pool. */ ASSERT(tx->tx_txg != TXG_INITIAL); ASSERT(SPA_VERSION_IS_SUPPORTED(version)); ASSERT(version >= spa_version(spa)); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_history_log_internal(spa, "set", tx, "version=%lld", (longlong_t)version); } /* * Set zpool properties. */ static void spa_sync_props(void *arg, dmu_tx_t *tx) { nvlist_t *nvp = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; objset_t *mos = spa->spa_meta_objset; nvpair_t *elem = NULL; mutex_enter(&spa->spa_props_lock); while ((elem = nvlist_next_nvpair(nvp, elem))) { uint64_t intval; char *strval, *fname; zpool_prop_t prop; const char *propname; zprop_type_t proptype; spa_feature_t fid; switch (prop = zpool_name_to_prop(nvpair_name(elem))) { case ZPOOL_PROP_INVAL: /* * We checked this earlier in spa_prop_validate(). */ ASSERT(zpool_prop_feature(nvpair_name(elem))); fname = strchr(nvpair_name(elem), '@') + 1; VERIFY0(zfeature_lookup_name(fname, &fid)); spa_feature_enable(spa, fid, tx); spa_history_log_internal(spa, "set", tx, "%s=enabled", nvpair_name(elem)); break; case ZPOOL_PROP_VERSION: intval = fnvpair_value_uint64(elem); /* * The version is synced separately before other * properties and should be correct by now. */ ASSERT3U(spa_version(spa), >=, intval); break; case ZPOOL_PROP_ALTROOT: /* * 'altroot' is a non-persistent property. It should * have been set temporarily at creation or import time. */ ASSERT(spa->spa_root != NULL); break; case ZPOOL_PROP_READONLY: case ZPOOL_PROP_CACHEFILE: /* * 'readonly' and 'cachefile' are also non-persistent * properties. */ break; case ZPOOL_PROP_COMMENT: strval = fnvpair_value_string(elem); if (spa->spa_comment != NULL) spa_strfree(spa->spa_comment); spa->spa_comment = spa_strdup(strval); /* * We need to dirty the configuration on all the vdevs * so that their labels get updated. We also need to * update the cache file to keep it in sync with the * MOS version. It's unnecessary to do this for pool * creation since the vdev's configuration has already * been dirtied. */ if (tx->tx_txg != TXG_INITIAL) { vdev_config_dirty(spa->spa_root_vdev); spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); break; case ZPOOL_PROP_COMPATIBILITY: strval = fnvpair_value_string(elem); if (spa->spa_compatibility != NULL) spa_strfree(spa->spa_compatibility); spa->spa_compatibility = spa_strdup(strval); /* * Dirty the configuration on vdevs as above. */ if (tx->tx_txg != TXG_INITIAL) { vdev_config_dirty(spa->spa_root_vdev); spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); break; default: /* * Set pool property values in the poolprops mos object. */ if (spa->spa_pool_props_object == 0) { spa->spa_pool_props_object = zap_create_link(mos, DMU_OT_POOL_PROPS, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_PROPS, tx); } /* normalize the property name */ propname = zpool_prop_to_name(prop); proptype = zpool_prop_get_type(prop); if (nvpair_type(elem) == DATA_TYPE_STRING) { ASSERT(proptype == PROP_TYPE_STRING); strval = fnvpair_value_string(elem); VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 1, strlen(strval) + 1, strval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); } else if (nvpair_type(elem) == DATA_TYPE_UINT64) { intval = fnvpair_value_uint64(elem); if (proptype == PROP_TYPE_INDEX) { const char *unused; VERIFY0(zpool_prop_index_to_string( prop, intval, &unused)); } VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 8, 1, &intval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%lld", nvpair_name(elem), (longlong_t)intval); } else { ASSERT(0); /* not allowed */ } switch (prop) { case ZPOOL_PROP_DELEGATION: spa->spa_delegation = intval; break; case ZPOOL_PROP_BOOTFS: spa->spa_bootfs = intval; break; case ZPOOL_PROP_FAILUREMODE: spa->spa_failmode = intval; break; case ZPOOL_PROP_AUTOTRIM: spa->spa_autotrim = intval; spa_async_request(spa, SPA_ASYNC_AUTOTRIM_RESTART); break; case ZPOOL_PROP_AUTOEXPAND: spa->spa_autoexpand = intval; if (tx->tx_txg != TXG_INITIAL) spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); break; case ZPOOL_PROP_MULTIHOST: spa->spa_multihost = intval; break; default: break; } } } mutex_exit(&spa->spa_props_lock); } /* * Perform one-time upgrade on-disk changes. spa_version() does not * reflect the new version this txg, so there must be no changes this * txg to anything that the upgrade code depends on after it executes. * Therefore this must be called after dsl_pool_sync() does the sync * tasks. */ static void spa_sync_upgrades(spa_t *spa, dmu_tx_t *tx) { if (spa_sync_pass(spa) != 1) return; dsl_pool_t *dp = spa->spa_dsl_pool; rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); if (spa->spa_ubsync.ub_version < SPA_VERSION_ORIGIN && spa->spa_uberblock.ub_version >= SPA_VERSION_ORIGIN) { dsl_pool_create_origin(dp, tx); /* Keeping the origin open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_NEXT_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_NEXT_CLONES) { dsl_pool_upgrade_clones(dp, tx); } if (spa->spa_ubsync.ub_version < SPA_VERSION_DIR_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_DIR_CLONES) { dsl_pool_upgrade_dir_clones(dp, tx); /* Keeping the freedir open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_FEATURES && spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) { spa_feature_create_zap_objects(spa, tx); } /* * LZ4_COMPRESS feature's behaviour was changed to activate_on_enable * when possibility to use lz4 compression for metadata was added * Old pools that have this feature enabled must be upgraded to have * this feature active */ if (spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) { boolean_t lz4_en = spa_feature_is_enabled(spa, SPA_FEATURE_LZ4_COMPRESS); boolean_t lz4_ac = spa_feature_is_active(spa, SPA_FEATURE_LZ4_COMPRESS); if (lz4_en && !lz4_ac) spa_feature_incr(spa, SPA_FEATURE_LZ4_COMPRESS, tx); } /* * If we haven't written the salt, do so now. Note that the * feature may not be activated yet, but that's fine since * the presence of this ZAP entry is backwards compatible. */ if (zap_contains(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CHECKSUM_SALT) == ENOENT) { VERIFY0(zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CHECKSUM_SALT, 1, sizeof (spa->spa_cksum_salt.zcs_bytes), spa->spa_cksum_salt.zcs_bytes, tx)); } rrw_exit(&dp->dp_config_rwlock, FTAG); } static void vdev_indirect_state_sync_verify(vdev_t *vd) { vdev_indirect_mapping_t *vim __maybe_unused = vd->vdev_indirect_mapping; vdev_indirect_births_t *vib __maybe_unused = vd->vdev_indirect_births; if (vd->vdev_ops == &vdev_indirect_ops) { ASSERT(vim != NULL); ASSERT(vib != NULL); } uint64_t obsolete_sm_object = 0; ASSERT0(vdev_obsolete_sm_object(vd, &obsolete_sm_object)); if (obsolete_sm_object != 0) { ASSERT(vd->vdev_obsolete_sm != NULL); ASSERT(vd->vdev_removing || vd->vdev_ops == &vdev_indirect_ops); ASSERT(vdev_indirect_mapping_num_entries(vim) > 0); ASSERT(vdev_indirect_mapping_bytes_mapped(vim) > 0); ASSERT3U(obsolete_sm_object, ==, space_map_object(vd->vdev_obsolete_sm)); ASSERT3U(vdev_indirect_mapping_bytes_mapped(vim), >=, space_map_allocated(vd->vdev_obsolete_sm)); } ASSERT(vd->vdev_obsolete_segments != NULL); /* * Since frees / remaps to an indirect vdev can only * happen in syncing context, the obsolete segments * tree must be empty when we start syncing. */ ASSERT0(range_tree_space(vd->vdev_obsolete_segments)); } /* * Set the top-level vdev's max queue depth. Evaluate each top-level's * async write queue depth in case it changed. The max queue depth will * not change in the middle of syncing out this txg. */ static void spa_sync_adjust_vdev_max_queue_depth(spa_t *spa) { ASSERT(spa_writeable(spa)); vdev_t *rvd = spa->spa_root_vdev; uint32_t max_queue_depth = zfs_vdev_async_write_max_active * zfs_vdev_queue_depth_pct / 100; metaslab_class_t *normal = spa_normal_class(spa); metaslab_class_t *special = spa_special_class(spa); metaslab_class_t *dedup = spa_dedup_class(spa); uint64_t slots_per_allocator = 0; for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (mg == NULL || !metaslab_group_initialized(mg)) continue; metaslab_class_t *mc = mg->mg_class; if (mc != normal && mc != special && mc != dedup) continue; /* * It is safe to do a lock-free check here because only async * allocations look at mg_max_alloc_queue_depth, and async * allocations all happen from spa_sync(). */ for (int i = 0; i < mg->mg_allocators; i++) { ASSERT0(zfs_refcount_count( &(mg->mg_allocator[i].mga_alloc_queue_depth))); } mg->mg_max_alloc_queue_depth = max_queue_depth; for (int i = 0; i < mg->mg_allocators; i++) { mg->mg_allocator[i].mga_cur_max_alloc_queue_depth = zfs_vdev_def_queue_depth; } slots_per_allocator += zfs_vdev_def_queue_depth; } for (int i = 0; i < spa->spa_alloc_count; i++) { ASSERT0(zfs_refcount_count(&normal->mc_allocator[i]. mca_alloc_slots)); ASSERT0(zfs_refcount_count(&special->mc_allocator[i]. mca_alloc_slots)); ASSERT0(zfs_refcount_count(&dedup->mc_allocator[i]. mca_alloc_slots)); normal->mc_allocator[i].mca_alloc_max_slots = slots_per_allocator; special->mc_allocator[i].mca_alloc_max_slots = slots_per_allocator; dedup->mc_allocator[i].mca_alloc_max_slots = slots_per_allocator; } normal->mc_alloc_throttle_enabled = zio_dva_throttle_enabled; special->mc_alloc_throttle_enabled = zio_dva_throttle_enabled; dedup->mc_alloc_throttle_enabled = zio_dva_throttle_enabled; } static void spa_sync_condense_indirect(spa_t *spa, dmu_tx_t *tx) { ASSERT(spa_writeable(spa)); vdev_t *rvd = spa->spa_root_vdev; for (int c = 0; c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; vdev_indirect_state_sync_verify(vd); if (vdev_indirect_should_condense(vd)) { spa_condense_indirect_start_sync(vd, tx); break; } } } static void spa_sync_iterate_to_convergence(spa_t *spa, dmu_tx_t *tx) { objset_t *mos = spa->spa_meta_objset; dsl_pool_t *dp = spa->spa_dsl_pool; uint64_t txg = tx->tx_txg; bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK]; do { int pass = ++spa->spa_sync_pass; spa_sync_config_object(spa, tx); spa_sync_aux_dev(spa, &spa->spa_spares, tx, ZPOOL_CONFIG_SPARES, DMU_POOL_SPARES); spa_sync_aux_dev(spa, &spa->spa_l2cache, tx, ZPOOL_CONFIG_L2CACHE, DMU_POOL_L2CACHE); spa_errlog_sync(spa, txg); dsl_pool_sync(dp, txg); if (pass < zfs_sync_pass_deferred_free || spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) { /* * If the log space map feature is active we don't * care about deferred frees and the deferred bpobj * as the log space map should effectively have the * same results (i.e. appending only to one object). */ spa_sync_frees(spa, free_bpl, tx); } else { /* * We can not defer frees in pass 1, because * we sync the deferred frees later in pass 1. */ ASSERT3U(pass, >, 1); bplist_iterate(free_bpl, bpobj_enqueue_alloc_cb, &spa->spa_deferred_bpobj, tx); } ddt_sync(spa, txg); dsl_scan_sync(dp, tx); svr_sync(spa, tx); spa_sync_upgrades(spa, tx); spa_flush_metaslabs(spa, tx); vdev_t *vd = NULL; while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg)) != NULL) vdev_sync(vd, txg); /* * Note: We need to check if the MOS is dirty because we could * have marked the MOS dirty without updating the uberblock * (e.g. if we have sync tasks but no dirty user data). We need * to check the uberblock's rootbp because it is updated if we * have synced out dirty data (though in this case the MOS will * most likely also be dirty due to second order effects, we * don't want to rely on that here). */ if (pass == 1 && spa->spa_uberblock.ub_rootbp.blk_birth < txg && !dmu_objset_is_dirty(mos, txg)) { /* * Nothing changed on the first pass, therefore this * TXG is a no-op. Avoid syncing deferred frees, so * that we can keep this TXG as a no-op. */ ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg)); ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg)); ASSERT(txg_list_empty(&dp->dp_sync_tasks, txg)); ASSERT(txg_list_empty(&dp->dp_early_sync_tasks, txg)); break; } spa_sync_deferred_frees(spa, tx); } while (dmu_objset_is_dirty(mos, txg)); } /* * Rewrite the vdev configuration (which includes the uberblock) to * commit the transaction group. * * If there are no dirty vdevs, we sync the uberblock to a few random * top-level vdevs that are known to be visible in the config cache * (see spa_vdev_add() for a complete description). If there *are* dirty * vdevs, sync the uberblock to all vdevs. */ static void spa_sync_rewrite_vdev_config(spa_t *spa, dmu_tx_t *tx) { vdev_t *rvd = spa->spa_root_vdev; uint64_t txg = tx->tx_txg; for (;;) { int error = 0; /* * We hold SCL_STATE to prevent vdev open/close/etc. * while we're attempting to write the vdev labels. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); if (list_is_empty(&spa->spa_config_dirty_list)) { vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL }; int svdcount = 0; int children = rvd->vdev_children; int c0 = random_in_range(children); for (int c = 0; c < children; c++) { vdev_t *vd = rvd->vdev_child[(c0 + c) % children]; /* Stop when revisiting the first vdev */ if (c > 0 && svd[0] == vd) break; if (vd->vdev_ms_array == 0 || vd->vdev_islog || !vdev_is_concrete(vd)) continue; svd[svdcount++] = vd; if (svdcount == SPA_SYNC_MIN_VDEVS) break; } error = vdev_config_sync(svd, svdcount, txg); } else { error = vdev_config_sync(rvd->vdev_child, rvd->vdev_children, txg); } if (error == 0) spa->spa_last_synced_guid = rvd->vdev_guid; spa_config_exit(spa, SCL_STATE, FTAG); if (error == 0) break; zio_suspend(spa, NULL, ZIO_SUSPEND_IOERR); zio_resume_wait(spa); } } /* * Sync the specified transaction group. New blocks may be dirtied as * part of the process, so we iterate until it converges. */ void spa_sync(spa_t *spa, uint64_t txg) { vdev_t *vd = NULL; VERIFY(spa_writeable(spa)); /* * Wait for i/os issued in open context that need to complete * before this txg syncs. */ (void) zio_wait(spa->spa_txg_zio[txg & TXG_MASK]); spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); /* * Lock out configuration changes. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa->spa_syncing_txg = txg; spa->spa_sync_pass = 0; for (int i = 0; i < spa->spa_alloc_count; i++) { mutex_enter(&spa->spa_allocs[i].spaa_lock); VERIFY0(avl_numnodes(&spa->spa_allocs[i].spaa_tree)); mutex_exit(&spa->spa_allocs[i].spaa_lock); } /* * If there are any pending vdev state changes, convert them * into config changes that go out with this transaction group. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); while (list_head(&spa->spa_state_dirty_list) != NULL) { /* * We need the write lock here because, for aux vdevs, * calling vdev_config_dirty() modifies sav_config. * This is ugly and will become unnecessary when we * eliminate the aux vdev wart by integrating all vdevs * into the root vdev tree. */ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_WRITER); while ((vd = list_head(&spa->spa_state_dirty_list)) != NULL) { vdev_state_clean(vd); vdev_config_dirty(vd); } spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } spa_config_exit(spa, SCL_STATE, FTAG); dsl_pool_t *dp = spa->spa_dsl_pool; dmu_tx_t *tx = dmu_tx_create_assigned(dp, txg); spa->spa_sync_starttime = gethrtime(); taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = taskq_dispatch_delay(system_delay_taskq, spa_deadman, spa, TQ_SLEEP, ddi_get_lbolt() + NSEC_TO_TICK(spa->spa_deadman_synctime)); /* * If we are upgrading to SPA_VERSION_RAIDZ_DEFLATE this txg, * set spa_deflate if we have no raid-z vdevs. */ if (spa->spa_ubsync.ub_version < SPA_VERSION_RAIDZ_DEFLATE && spa->spa_uberblock.ub_version >= SPA_VERSION_RAIDZ_DEFLATE) { vdev_t *rvd = spa->spa_root_vdev; int i; for (i = 0; i < rvd->vdev_children; i++) { vd = rvd->vdev_child[i]; if (vd->vdev_deflate_ratio != SPA_MINBLOCKSIZE) break; } if (i == rvd->vdev_children) { spa->spa_deflate = TRUE; VERIFY0(zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx)); } } spa_sync_adjust_vdev_max_queue_depth(spa); spa_sync_condense_indirect(spa, tx); spa_sync_iterate_to_convergence(spa, tx); #ifdef ZFS_DEBUG if (!list_is_empty(&spa->spa_config_dirty_list)) { /* * Make sure that the number of ZAPs for all the vdevs matches * the number of ZAPs in the per-vdev ZAP list. This only gets * called if the config is dirty; otherwise there may be * outstanding AVZ operations that weren't completed in * spa_sync_config_object. */ uint64_t all_vdev_zap_entry_count; ASSERT0(zap_count(spa->spa_meta_objset, spa->spa_all_vdev_zaps, &all_vdev_zap_entry_count)); ASSERT3U(vdev_count_verify_zaps(spa->spa_root_vdev), ==, all_vdev_zap_entry_count); } #endif if (spa->spa_vdev_removal != NULL) { ASSERT0(spa->spa_vdev_removal->svr_bytes_done[txg & TXG_MASK]); } spa_sync_rewrite_vdev_config(spa, tx); dmu_tx_commit(tx); taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = 0; /* * Clear the dirty config list. */ while ((vd = list_head(&spa->spa_config_dirty_list)) != NULL) vdev_config_clean(vd); /* * Now that the new config has synced transactionally, * let it become visible to the config cache. */ if (spa->spa_config_syncing != NULL) { spa_config_set(spa, spa->spa_config_syncing); spa->spa_config_txg = txg; spa->spa_config_syncing = NULL; } dsl_pool_sync_done(dp, txg); for (int i = 0; i < spa->spa_alloc_count; i++) { mutex_enter(&spa->spa_allocs[i].spaa_lock); VERIFY0(avl_numnodes(&spa->spa_allocs[i].spaa_tree)); mutex_exit(&spa->spa_allocs[i].spaa_lock); } /* * Update usable space statistics. */ while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))) != NULL) vdev_sync_done(vd, txg); metaslab_class_evict_old(spa->spa_normal_class, txg); metaslab_class_evict_old(spa->spa_log_class, txg); spa_sync_close_syncing_log_sm(spa); spa_update_dspace(spa); /* * It had better be the case that we didn't dirty anything * since vdev_config_sync(). */ ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg)); ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg)); ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg)); while (zfs_pause_spa_sync) delay(1); spa->spa_sync_pass = 0; /* * Update the last synced uberblock here. We want to do this at * the end of spa_sync() so that consumers of spa_last_synced_txg() * will be guaranteed that all the processing associated with * that txg has been completed. */ spa->spa_ubsync = spa->spa_uberblock; spa_config_exit(spa, SCL_CONFIG, FTAG); spa_handle_ignored_writes(spa); /* * If any async tasks have been requested, kick them off. */ spa_async_dispatch(spa); } /* * Sync all pools. We don't want to hold the namespace lock across these * operations, so we take a reference on the spa_t and drop the lock during the * sync. */ void spa_sync_allpools(void) { spa_t *spa = NULL; mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) { if (spa_state(spa) != POOL_STATE_ACTIVE || !spa_writeable(spa) || spa_suspended(spa)) continue; spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); txg_wait_synced(spa_get_dsl(spa), 0); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); } mutex_exit(&spa_namespace_lock); } /* * ========================================================================== * Miscellaneous routines * ========================================================================== */ /* * Remove all pools in the system. */ void spa_evict_all(void) { spa_t *spa; /* * Remove all cached state. All pools should be closed now, * so every spa in the AVL tree should be unreferenced. */ mutex_enter(&spa_namespace_lock); while ((spa = spa_next(NULL)) != NULL) { /* * Stop async tasks. The async thread may need to detach * a device that's been replaced, which requires grabbing * spa_namespace_lock, so we must drop it here. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } spa_remove(spa); } mutex_exit(&spa_namespace_lock); } vdev_t * spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t aux) { vdev_t *vd; int i; if ((vd = vdev_lookup_by_guid(spa->spa_root_vdev, guid)) != NULL) return (vd); if (aux) { for (i = 0; i < spa->spa_l2cache.sav_count; i++) { vd = spa->spa_l2cache.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } } return (NULL); } void spa_upgrade(spa_t *spa, uint64_t version) { ASSERT(spa_writeable(spa)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * This should only be called for a non-faulted pool, and since a * future version would result in an unopenable pool, this shouldn't be * possible. */ ASSERT(SPA_VERSION_IS_SUPPORTED(spa->spa_uberblock.ub_version)); ASSERT3U(version, >=, spa->spa_uberblock.ub_version); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); txg_wait_synced(spa_get_dsl(spa), 0); } static boolean_t spa_has_aux_vdev(spa_t *spa, uint64_t guid, spa_aux_vdev_t *sav) { (void) spa; int i; uint64_t vdev_guid; for (i = 0; i < sav->sav_count; i++) if (sav->sav_vdevs[i]->vdev_guid == guid) return (B_TRUE); for (i = 0; i < sav->sav_npending; i++) { if (nvlist_lookup_uint64(sav->sav_pending[i], ZPOOL_CONFIG_GUID, &vdev_guid) == 0 && vdev_guid == guid) return (B_TRUE); } return (B_FALSE); } boolean_t spa_has_l2cache(spa_t *spa, uint64_t guid) { return (spa_has_aux_vdev(spa, guid, &spa->spa_l2cache)); } boolean_t spa_has_spare(spa_t *spa, uint64_t guid) { return (spa_has_aux_vdev(spa, guid, &spa->spa_spares)); } /* * Check if a pool has an active shared spare device. * Note: reference count of an active spare is 2, as a spare and as a replace */ static boolean_t spa_has_active_shared_spare(spa_t *spa) { int i, refcnt; uint64_t pool; spa_aux_vdev_t *sav = &spa->spa_spares; for (i = 0; i < sav->sav_count; i++) { if (spa_spare_exists(sav->sav_vdevs[i]->vdev_guid, &pool, &refcnt) && pool != 0ULL && pool == spa_guid(spa) && refcnt > 2) return (B_TRUE); } return (B_FALSE); } uint64_t spa_total_metaslabs(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; uint64_t m = 0; for (uint64_t c = 0; c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; if (!vdev_is_concrete(vd)) continue; m += vd->vdev_ms_count; } return (m); } /* * Notify any waiting threads that some activity has switched from being in- * progress to not-in-progress so that the thread can wake up and determine * whether it is finished waiting. */ void spa_notify_waiters(spa_t *spa) { /* * Acquiring spa_activities_lock here prevents the cv_broadcast from * happening between the waiting thread's check and cv_wait. */ mutex_enter(&spa->spa_activities_lock); cv_broadcast(&spa->spa_activities_cv); mutex_exit(&spa->spa_activities_lock); } /* * Notify any waiting threads that the pool is exporting, and then block until * they are finished using the spa_t. */ void spa_wake_waiters(spa_t *spa) { mutex_enter(&spa->spa_activities_lock); spa->spa_waiters_cancel = B_TRUE; cv_broadcast(&spa->spa_activities_cv); while (spa->spa_waiters != 0) cv_wait(&spa->spa_waiters_cv, &spa->spa_activities_lock); spa->spa_waiters_cancel = B_FALSE; mutex_exit(&spa->spa_activities_lock); } /* Whether the vdev or any of its descendants are being initialized/trimmed. */ static boolean_t spa_vdev_activity_in_progress_impl(vdev_t *vd, zpool_wait_activity_t activity) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER)); ASSERT(MUTEX_HELD(&spa->spa_activities_lock)); ASSERT(activity == ZPOOL_WAIT_INITIALIZE || activity == ZPOOL_WAIT_TRIM); kmutex_t *lock = activity == ZPOOL_WAIT_INITIALIZE ? &vd->vdev_initialize_lock : &vd->vdev_trim_lock; mutex_exit(&spa->spa_activities_lock); mutex_enter(lock); mutex_enter(&spa->spa_activities_lock); boolean_t in_progress = (activity == ZPOOL_WAIT_INITIALIZE) ? (vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) : (vd->vdev_trim_state == VDEV_TRIM_ACTIVE); mutex_exit(lock); if (in_progress) return (B_TRUE); for (int i = 0; i < vd->vdev_children; i++) { if (spa_vdev_activity_in_progress_impl(vd->vdev_child[i], activity)) return (B_TRUE); } return (B_FALSE); } /* * If use_guid is true, this checks whether the vdev specified by guid is * being initialized/trimmed. Otherwise, it checks whether any vdev in the pool * is being initialized/trimmed. The caller must hold the config lock and * spa_activities_lock. */ static int spa_vdev_activity_in_progress(spa_t *spa, boolean_t use_guid, uint64_t guid, zpool_wait_activity_t activity, boolean_t *in_progress) { mutex_exit(&spa->spa_activities_lock); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); mutex_enter(&spa->spa_activities_lock); vdev_t *vd; if (use_guid) { vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd == NULL || !vd->vdev_ops->vdev_op_leaf) { spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (EINVAL); } } else { vd = spa->spa_root_vdev; } *in_progress = spa_vdev_activity_in_progress_impl(vd, activity); spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (0); } /* * Locking for waiting threads * --------------------------- * * Waiting threads need a way to check whether a given activity is in progress, * and then, if it is, wait for it to complete. Each activity will have some * in-memory representation of the relevant on-disk state which can be used to * determine whether or not the activity is in progress. The in-memory state and * the locking used to protect it will be different for each activity, and may * not be suitable for use with a cvar (e.g., some state is protected by the * config lock). To allow waiting threads to wait without any races, another * lock, spa_activities_lock, is used. * * When the state is checked, both the activity-specific lock (if there is one) * and spa_activities_lock are held. In some cases, the activity-specific lock * is acquired explicitly (e.g. the config lock). In others, the locking is * internal to some check (e.g. bpobj_is_empty). After checking, the waiting * thread releases the activity-specific lock and, if the activity is in * progress, then cv_waits using spa_activities_lock. * * The waiting thread is woken when another thread, one completing some * activity, updates the state of the activity and then calls * spa_notify_waiters, which will cv_broadcast. This 'completing' thread only * needs to hold its activity-specific lock when updating the state, and this * lock can (but doesn't have to) be dropped before calling spa_notify_waiters. * * Because spa_notify_waiters acquires spa_activities_lock before broadcasting, * and because it is held when the waiting thread checks the state of the * activity, it can never be the case that the completing thread both updates * the activity state and cv_broadcasts in between the waiting thread's check * and cv_wait. Thus, a waiting thread can never miss a wakeup. * * In order to prevent deadlock, when the waiting thread does its check, in some * cases it will temporarily drop spa_activities_lock in order to acquire the * activity-specific lock. The order in which spa_activities_lock and the * activity specific lock are acquired in the waiting thread is determined by * the order in which they are acquired in the completing thread; if the * completing thread calls spa_notify_waiters with the activity-specific lock * held, then the waiting thread must also acquire the activity-specific lock * first. */ static int spa_activity_in_progress(spa_t *spa, zpool_wait_activity_t activity, boolean_t use_tag, uint64_t tag, boolean_t *in_progress) { int error = 0; ASSERT(MUTEX_HELD(&spa->spa_activities_lock)); switch (activity) { case ZPOOL_WAIT_CKPT_DISCARD: *in_progress = (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT) && zap_contains(spa_meta_objset(spa), DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT) == ENOENT); break; case ZPOOL_WAIT_FREE: *in_progress = ((spa_version(spa) >= SPA_VERSION_DEADLISTS && !bpobj_is_empty(&spa->spa_dsl_pool->dp_free_bpobj)) || spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY) || spa_livelist_delete_check(spa)); break; case ZPOOL_WAIT_INITIALIZE: case ZPOOL_WAIT_TRIM: error = spa_vdev_activity_in_progress(spa, use_tag, tag, activity, in_progress); break; case ZPOOL_WAIT_REPLACE: mutex_exit(&spa->spa_activities_lock); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); mutex_enter(&spa->spa_activities_lock); *in_progress = vdev_replace_in_progress(spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); break; case ZPOOL_WAIT_REMOVE: *in_progress = (spa->spa_removing_phys.sr_state == DSS_SCANNING); break; case ZPOOL_WAIT_RESILVER: if ((*in_progress = vdev_rebuild_active(spa->spa_root_vdev))) break; zfs_fallthrough; case ZPOOL_WAIT_SCRUB: { boolean_t scanning, paused, is_scrub; dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; is_scrub = (scn->scn_phys.scn_func == POOL_SCAN_SCRUB); scanning = (scn->scn_phys.scn_state == DSS_SCANNING); paused = dsl_scan_is_paused_scrub(scn); *in_progress = (scanning && !paused && is_scrub == (activity == ZPOOL_WAIT_SCRUB)); break; } default: panic("unrecognized value for activity %d", activity); } return (error); } static int spa_wait_common(const char *pool, zpool_wait_activity_t activity, boolean_t use_tag, uint64_t tag, boolean_t *waited) { /* * The tag is used to distinguish between instances of an activity. * 'initialize' and 'trim' are the only activities that we use this for. * The other activities can only have a single instance in progress in a * pool at one time, making the tag unnecessary. * * There can be multiple devices being replaced at once, but since they * all finish once resilvering finishes, we don't bother keeping track * of them individually, we just wait for them all to finish. */ if (use_tag && activity != ZPOOL_WAIT_INITIALIZE && activity != ZPOOL_WAIT_TRIM) return (EINVAL); if (activity < 0 || activity >= ZPOOL_WAIT_NUM_ACTIVITIES) return (EINVAL); spa_t *spa; int error = spa_open(pool, &spa, FTAG); if (error != 0) return (error); /* * Increment the spa's waiter count so that we can call spa_close and * still ensure that the spa_t doesn't get freed before this thread is * finished with it when the pool is exported. We want to call spa_close * before we start waiting because otherwise the additional ref would * prevent the pool from being exported or destroyed throughout the * potentially long wait. */ mutex_enter(&spa->spa_activities_lock); spa->spa_waiters++; spa_close(spa, FTAG); *waited = B_FALSE; for (;;) { boolean_t in_progress; error = spa_activity_in_progress(spa, activity, use_tag, tag, &in_progress); if (error || !in_progress || spa->spa_waiters_cancel) break; *waited = B_TRUE; if (cv_wait_sig(&spa->spa_activities_cv, &spa->spa_activities_lock) == 0) { error = EINTR; break; } } spa->spa_waiters--; cv_signal(&spa->spa_waiters_cv); mutex_exit(&spa->spa_activities_lock); return (error); } /* * Wait for a particular instance of the specified activity to complete, where * the instance is identified by 'tag' */ int spa_wait_tag(const char *pool, zpool_wait_activity_t activity, uint64_t tag, boolean_t *waited) { return (spa_wait_common(pool, activity, B_TRUE, tag, waited)); } /* * Wait for all instances of the specified activity complete */ int spa_wait(const char *pool, zpool_wait_activity_t activity, boolean_t *waited) { return (spa_wait_common(pool, activity, B_FALSE, 0, waited)); } sysevent_t * spa_event_create(spa_t *spa, vdev_t *vd, nvlist_t *hist_nvl, const char *name) { sysevent_t *ev = NULL; #ifdef _KERNEL nvlist_t *resource; resource = zfs_event_create(spa, vd, FM_SYSEVENT_CLASS, name, hist_nvl); if (resource) { ev = kmem_alloc(sizeof (sysevent_t), KM_SLEEP); ev->resource = resource; } #else (void) spa, (void) vd, (void) hist_nvl, (void) name; #endif return (ev); } void spa_event_post(sysevent_t *ev) { #ifdef _KERNEL if (ev) { zfs_zevent_post(ev->resource, NULL, zfs_zevent_post_cb); kmem_free(ev, sizeof (*ev)); } #else (void) ev; #endif } /* * Post a zevent corresponding to the given sysevent. The 'name' must be one * of the event definitions in sys/sysevent/eventdefs.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. */ void spa_event_notify(spa_t *spa, vdev_t *vd, nvlist_t *hist_nvl, const char *name) { spa_event_post(spa_event_create(spa, vd, hist_nvl, name)); } /* 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); 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 manipulation */ EXPORT_SYMBOL(spa_vdev_add); EXPORT_SYMBOL(spa_vdev_attach); EXPORT_SYMBOL(spa_vdev_detach); 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); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_shift, UINT, ZMOD_RW, "log2 fraction of arc that can be used by inflight I/Os when " "verifying pool during import"); /* END CSTYLED */ ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_metadata, INT, ZMOD_RW, "Set to traverse metadata on pool import"); ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_data, INT, ZMOD_RW, "Set to traverse data on pool import"); ZFS_MODULE_PARAM(zfs_spa, spa_, load_print_vdev_tree, INT, ZMOD_RW, "Print vdev tree to zfs_dbgmsg during pool import"); ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_pct, UINT, ZMOD_RD, "Percentage of CPUs to run an IO worker thread"); ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_tpq, UINT, ZMOD_RD, "Number of threads per IO worker taskqueue"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs, zfs_, max_missing_tvds, ULONG, ZMOD_RW, "Allow importing pool with up to this number of missing top-level " "vdevs (in read-only mode)"); /* END CSTYLED */ ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, zthr_pause, INT, ZMOD_RW, "Set the livelist condense zthr to pause"); ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, sync_pause, INT, ZMOD_RW, "Set the livelist condense synctask to pause"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, sync_cancel, INT, ZMOD_RW, "Whether livelist condensing was canceled in the synctask"); ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, zthr_cancel, INT, ZMOD_RW, "Whether livelist condensing was canceled in the zthr function"); ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, new_alloc, INT, ZMOD_RW, "Whether extra ALLOC blkptrs were added to a livelist entry while it " "was being condensed"); /* END CSTYLED */ diff --git a/module/zfs/spa_config.c b/module/zfs/spa_config.c index 91ac5c05e8af..5165c370403b 100644 --- a/module/zfs/spa_config.c +++ b/module/zfs/spa_config.c @@ -1,623 +1,635 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright 2017 Joyent, Inc. * Copyright (c) 2021, Colm Buckley */ #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif /* * Pool configuration repository. * * Pool configuration is stored as a packed nvlist on the filesystem. By * default, all pools are stored in /etc/zfs/zpool.cache and loaded on boot * (when the ZFS module is loaded). Pools can also have the 'cachefile' * property set that allows them to be stored in an alternate location until * the control of external software. * * For each cache file, we have a single nvlist which holds all the * configuration information. When the module loads, we read this information * from /etc/zfs/zpool.cache and populate the SPA namespace. This namespace is * maintained independently in spa.c. Whenever the namespace is modified, or * the configuration of a pool is changed, we call spa_write_cachefile(), which * walks through all the active pools and writes the configuration to disk. */ static uint64_t spa_config_generation = 1; /* * This can be overridden in userland to preserve an alternate namespace for * userland pools when doing testing. */ char *spa_config_path = (char *)ZPOOL_CACHE; #ifdef _KERNEL static int zfs_autoimport_disable = B_TRUE; #endif /* * Called when the module is first loaded, this routine loads the configuration * file into the SPA namespace. It does not actually open or load the pools; it * only populates the namespace. */ void spa_config_load(void) { void *buf = NULL; nvlist_t *nvlist, *child; nvpair_t *nvpair; char *pathname; zfs_file_t *fp; zfs_file_attr_t zfa; uint64_t fsize; int err; #ifdef _KERNEL if (zfs_autoimport_disable) return; #endif /* * Open the configuration file. */ pathname = kmem_alloc(MAXPATHLEN, KM_SLEEP); (void) snprintf(pathname, MAXPATHLEN, "%s", spa_config_path); err = zfs_file_open(pathname, O_RDONLY, 0, &fp); #ifdef __FreeBSD__ if (err) err = zfs_file_open(ZPOOL_CACHE_BOOT, O_RDONLY, 0, &fp); #endif kmem_free(pathname, MAXPATHLEN); if (err) return; if (zfs_file_getattr(fp, &zfa)) goto out; fsize = zfa.zfa_size; buf = kmem_alloc(fsize, KM_SLEEP); /* * Read the nvlist from the file. */ if (zfs_file_read(fp, buf, fsize, NULL) < 0) goto out; /* * Unpack the nvlist. */ if (nvlist_unpack(buf, fsize, &nvlist, KM_SLEEP) != 0) goto out; /* * Iterate over all elements in the nvlist, creating a new spa_t for * each one with the specified configuration. */ mutex_enter(&spa_namespace_lock); nvpair = NULL; while ((nvpair = nvlist_next_nvpair(nvlist, nvpair)) != NULL) { if (nvpair_type(nvpair) != DATA_TYPE_NVLIST) continue; child = fnvpair_value_nvlist(nvpair); if (spa_lookup(nvpair_name(nvpair)) != NULL) continue; (void) spa_add(nvpair_name(nvpair), child, NULL); } mutex_exit(&spa_namespace_lock); nvlist_free(nvlist); out: if (buf != NULL) kmem_free(buf, fsize); zfs_file_close(fp); } static int spa_config_remove(spa_config_dirent_t *dp) { int error = 0; /* * Remove the cache file. If zfs_file_unlink() in not supported by the * platform fallback to truncating the file which is functionally * equivalent. */ error = zfs_file_unlink(dp->scd_path); if (error == EOPNOTSUPP) { int flags = O_RDWR | O_TRUNC; zfs_file_t *fp; error = zfs_file_open(dp->scd_path, flags, 0644, &fp); if (error == 0) { (void) zfs_file_fsync(fp, O_SYNC); (void) zfs_file_close(fp); } } return (error); } static int spa_config_write(spa_config_dirent_t *dp, nvlist_t *nvl) { size_t buflen; char *buf; int oflags = O_RDWR | O_TRUNC | O_CREAT | O_LARGEFILE; char *temp; int err; zfs_file_t *fp; /* * If the nvlist is empty (NULL), then remove the old cachefile. */ if (nvl == NULL) { err = spa_config_remove(dp); if (err == ENOENT) err = 0; return (err); } /* * Pack the configuration into a buffer. */ buf = fnvlist_pack(nvl, &buflen); temp = kmem_zalloc(MAXPATHLEN, KM_SLEEP); /* * Write the configuration to disk. Due to the complexity involved * in performing a rename and remove from within the kernel the file * is instead truncated and overwritten in place. This way we always * have a consistent view of the data or a zero length file. */ err = zfs_file_open(dp->scd_path, oflags, 0644, &fp); if (err == 0) { err = zfs_file_write(fp, buf, buflen, NULL); if (err == 0) err = zfs_file_fsync(fp, O_SYNC); zfs_file_close(fp); if (err) (void) spa_config_remove(dp); } fnvlist_pack_free(buf, buflen); kmem_free(temp, MAXPATHLEN); return (err); } /* * Synchronize pool configuration to disk. This must be called with the * namespace lock held. Synchronizing the pool cache is typically done after * the configuration has been synced to the MOS. This exposes a window where * the MOS config will have been updated but the cache file has not. If * the system were to crash at that instant then the cached config may not * contain the correct information to open the pool and an explicit import * would be required. */ void -spa_write_cachefile(spa_t *target, boolean_t removing, boolean_t postsysevent) +spa_write_cachefile(spa_t *target, boolean_t removing, boolean_t postsysevent, + boolean_t postblkidevent) { spa_config_dirent_t *dp, *tdp; nvlist_t *nvl; char *pool_name; boolean_t ccw_failure; int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (!(spa_mode_global & SPA_MODE_WRITE)) return; /* * Iterate over all cachefiles for the pool, past or present. When the * cachefile is changed, the new one is pushed onto this list, allowing * us to update previous cachefiles that no longer contain this pool. */ ccw_failure = B_FALSE; for (dp = list_head(&target->spa_config_list); dp != NULL; dp = list_next(&target->spa_config_list, dp)) { spa_t *spa = NULL; if (dp->scd_path == NULL) continue; /* * Iterate over all pools, adding any matching pools to 'nvl'. */ nvl = NULL; while ((spa = spa_next(spa)) != NULL) { /* * Skip over our own pool if we're about to remove * ourselves from the spa namespace or any pool that * is readonly. Since we cannot guarantee that a * readonly pool would successfully import upon reboot, * we don't allow them to be written to the cache file. */ if ((spa == target && removing) || !spa_writeable(spa)) continue; mutex_enter(&spa->spa_props_lock); tdp = list_head(&spa->spa_config_list); if (spa->spa_config == NULL || tdp == NULL || tdp->scd_path == NULL || strcmp(tdp->scd_path, dp->scd_path) != 0) { mutex_exit(&spa->spa_props_lock); continue; } if (nvl == NULL) nvl = fnvlist_alloc(); if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) pool_name = fnvlist_lookup_string( spa->spa_config, ZPOOL_CONFIG_POOL_NAME); else pool_name = spa_name(spa); fnvlist_add_nvlist(nvl, pool_name, spa->spa_config); mutex_exit(&spa->spa_props_lock); } error = spa_config_write(dp, nvl); if (error != 0) ccw_failure = B_TRUE; nvlist_free(nvl); } if (ccw_failure) { /* * Keep trying so that configuration data is * written if/when any temporary filesystem * resource issues are resolved. */ if (target->spa_ccw_fail_time == 0) { (void) zfs_ereport_post( FM_EREPORT_ZFS_CONFIG_CACHE_WRITE, target, NULL, NULL, NULL, 0); } target->spa_ccw_fail_time = gethrtime(); spa_async_request(target, SPA_ASYNC_CONFIG_UPDATE); } else { /* * Do not rate limit future attempts to update * the config cache. */ target->spa_ccw_fail_time = 0; } /* * Remove any config entries older than the current one. */ dp = list_head(&target->spa_config_list); while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) { list_remove(&target->spa_config_list, tdp); if (tdp->scd_path != NULL) spa_strfree(tdp->scd_path); kmem_free(tdp, sizeof (spa_config_dirent_t)); } spa_config_generation++; if (postsysevent) spa_event_notify(target, NULL, NULL, ESC_ZFS_CONFIG_SYNC); + + /* + * Post udev event to sync blkid information if the pool is created + * or a new vdev is added to the pool. + */ + if ((target->spa_root_vdev) && postblkidevent) { + vdev_post_kobj_evt(target->spa_root_vdev); + for (int i = 0; i < target->spa_l2cache.sav_count; i++) + vdev_post_kobj_evt(target->spa_l2cache.sav_vdevs[i]); + } } /* * Sigh. Inside a local zone, we don't have access to /etc/zfs/zpool.cache, * and we don't want to allow the local zone to see all the pools anyway. * So we have to invent the ZFS_IOC_CONFIG ioctl to grab the configuration * information for all pool visible within the zone. */ nvlist_t * spa_all_configs(uint64_t *generation) { nvlist_t *pools; spa_t *spa = NULL; if (*generation == spa_config_generation) return (NULL); pools = fnvlist_alloc(); mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) { if (INGLOBALZONE(curproc) || zone_dataset_visible(spa_name(spa), NULL)) { mutex_enter(&spa->spa_props_lock); fnvlist_add_nvlist(pools, spa_name(spa), spa->spa_config); mutex_exit(&spa->spa_props_lock); } } *generation = spa_config_generation; mutex_exit(&spa_namespace_lock); return (pools); } void spa_config_set(spa_t *spa, nvlist_t *config) { mutex_enter(&spa->spa_props_lock); if (spa->spa_config != NULL && spa->spa_config != config) nvlist_free(spa->spa_config); spa->spa_config = config; mutex_exit(&spa->spa_props_lock); } /* * Generate the pool's configuration based on the current in-core state. * * We infer whether to generate a complete config or just one top-level config * based on whether vd is the root vdev. */ nvlist_t * spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats) { nvlist_t *config, *nvroot; vdev_t *rvd = spa->spa_root_vdev; unsigned long hostid = 0; boolean_t locked = B_FALSE; uint64_t split_guid; char *pool_name; if (vd == NULL) { vd = rvd; locked = B_TRUE; spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) == (SCL_CONFIG | SCL_STATE)); /* * If txg is -1, report the current value of spa->spa_config_txg. */ if (txg == -1ULL) txg = spa->spa_config_txg; /* * Originally, users had to handle spa namespace collisions by either * exporting the already imported pool or by specifying a new name for * the pool with a conflicting name. In the case of root pools from * virtual guests, neither approach to collision resolution is * reasonable. This is addressed by extending the new name syntax with * an option to specify that the new name is temporary. When specified, * ZFS_IMPORT_TEMP_NAME will be set in spa->spa_import_flags to tell us * to use the previous name, which we do below. */ if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) { VERIFY0(nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &pool_name)); } else pool_name = spa_name(spa); config = fnvlist_alloc(); fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)); fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, pool_name); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, spa_state(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, txg); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata); if (spa->spa_comment != NULL) fnvlist_add_string(config, ZPOOL_CONFIG_COMMENT, spa->spa_comment); if (spa->spa_compatibility != NULL) fnvlist_add_string(config, ZPOOL_CONFIG_COMPATIBILITY, spa->spa_compatibility); hostid = spa_get_hostid(spa); if (hostid != 0) fnvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID, hostid); fnvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME, utsname()->nodename); int config_gen_flags = 0; if (vd != rvd) { fnvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID, vd->vdev_top->vdev_guid); fnvlist_add_uint64(config, ZPOOL_CONFIG_GUID, vd->vdev_guid); if (vd->vdev_isspare) fnvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE, 1ULL); if (vd->vdev_islog) fnvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG, 1ULL); vd = vd->vdev_top; /* label contains top config */ } else { /* * Only add the (potentially large) split information * in the mos config, and not in the vdev labels */ if (spa->spa_config_splitting != NULL) fnvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT, spa->spa_config_splitting); fnvlist_add_boolean(config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS); config_gen_flags |= VDEV_CONFIG_MOS; } /* * Add the top-level config. We even add this on pools which * don't support holes in the namespace. */ vdev_top_config_generate(spa, config); /* * If we're splitting, record the original pool's guid. */ if (spa->spa_config_splitting != NULL && nvlist_lookup_uint64(spa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) { fnvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID, split_guid); } nvroot = vdev_config_generate(spa, vd, getstats, config_gen_flags); fnvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot); nvlist_free(nvroot); /* * Store what's necessary for reading the MOS in the label. */ fnvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, spa->spa_label_features); if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) { ddt_histogram_t *ddh; ddt_stat_t *dds; ddt_object_t *ddo; ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_SLEEP); ddt_get_dedup_histogram(spa, ddh); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_HISTOGRAM, (uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)); kmem_free(ddh, sizeof (ddt_histogram_t)); ddo = kmem_zalloc(sizeof (ddt_object_t), KM_SLEEP); ddt_get_dedup_object_stats(spa, ddo); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_OBJ_STATS, (uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)); kmem_free(ddo, sizeof (ddt_object_t)); dds = kmem_zalloc(sizeof (ddt_stat_t), KM_SLEEP); ddt_get_dedup_stats(spa, dds); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_STATS, (uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)); kmem_free(dds, sizeof (ddt_stat_t)); } if (locked) spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (config); } /* * Update all disk labels, generate a fresh config based on the current * in-core state, and sync the global config cache (do not sync the config * cache if this is a booting rootpool). */ void spa_config_update(spa_t *spa, int what) { vdev_t *rvd = spa->spa_root_vdev; uint64_t txg; int c; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); txg = spa_last_synced_txg(spa) + 1; if (what == SPA_CONFIG_UPDATE_POOL) { vdev_config_dirty(rvd); } else { /* * If we have top-level vdevs that were added but have * not yet been prepared for allocation, do that now. * (It's safe now because the config cache is up to date, * so it will be able to translate the new DVAs.) * See comments in spa_vdev_add() for full details. */ for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; /* * Explicitly skip vdevs that are indirect or * log vdevs that are being removed. The reason * is that both of those can have vdev_ms_array * set to 0 and we wouldn't want to change their * metaslab size nor call vdev_expand() on them. */ if (!vdev_is_concrete(tvd) || (tvd->vdev_islog && tvd->vdev_removing)) continue; if (tvd->vdev_ms_array == 0) vdev_metaslab_set_size(tvd); vdev_expand(tvd, txg); } } spa_config_exit(spa, SCL_ALL, FTAG); /* * Wait for the mosconfig to be regenerated and synced. */ txg_wait_synced(spa->spa_dsl_pool, txg); /* * Update the global config cache to reflect the new mosconfig. */ if (!spa->spa_is_root) { spa_write_cachefile(spa, B_FALSE, + what != SPA_CONFIG_UPDATE_POOL, what != SPA_CONFIG_UPDATE_POOL); } if (what == SPA_CONFIG_UPDATE_POOL) spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS); } EXPORT_SYMBOL(spa_config_load); EXPORT_SYMBOL(spa_all_configs); EXPORT_SYMBOL(spa_config_set); EXPORT_SYMBOL(spa_config_generate); EXPORT_SYMBOL(spa_config_update); #ifdef __linux__ /* string sysctls require a char array on FreeBSD */ ZFS_MODULE_PARAM(zfs_spa, spa_, config_path, STRING, ZMOD_RD, "SPA config file (/etc/zfs/zpool.cache)"); #endif ZFS_MODULE_PARAM(zfs, zfs_, autoimport_disable, INT, ZMOD_RW, "Disable pool import at module load"); diff --git a/module/zfs/spa_misc.c b/module/zfs/spa_misc.c index f7865bc49bc4..102070013404 100644 --- a/module/zfs/spa_misc.c +++ b/module/zfs/spa_misc.c @@ -1,2959 +1,2959 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2019 by Delphix. All rights reserved. * Copyright 2015 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2017 Datto Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, loli10K . All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_prop.h" #include #include #include #include /* * SPA locking * * There are three basic locks for managing spa_t structures: * * spa_namespace_lock (global mutex) * * This lock must be acquired to do any of the following: * * - Lookup a spa_t by name * - Add or remove a spa_t from the namespace * - Increase spa_refcount from non-zero * - Check if spa_refcount is zero * - Rename a spa_t * - add/remove/attach/detach devices * - Held for the duration of create/destroy/import/export * * It does not need to handle recursion. A create or destroy may * reference objects (files or zvols) in other pools, but by * definition they must have an existing reference, and will never need * to lookup a spa_t by name. * * spa_refcount (per-spa zfs_refcount_t protected by mutex) * * This reference count keep track of any active users of the spa_t. The * spa_t cannot be destroyed or freed while this is non-zero. Internally, * the refcount is never really 'zero' - opening a pool implicitly keeps * some references in the DMU. Internally we check against spa_minref, but * present the image of a zero/non-zero value to consumers. * * spa_config_lock[] (per-spa array of rwlocks) * * This protects the spa_t from config changes, and must be held in * the following circumstances: * * - RW_READER to perform I/O to the spa * - RW_WRITER to change the vdev config * * The locking order is fairly straightforward: * * spa_namespace_lock -> spa_refcount * * The namespace lock must be acquired to increase the refcount from 0 * or to check if it is zero. * * spa_refcount -> spa_config_lock[] * * There must be at least one valid reference on the spa_t to acquire * the config lock. * * spa_namespace_lock -> spa_config_lock[] * * The namespace lock must always be taken before the config lock. * * * The spa_namespace_lock can be acquired directly and is globally visible. * * The namespace is manipulated using the following functions, all of which * require the spa_namespace_lock to be held. * * spa_lookup() Lookup a spa_t by name. * * spa_add() Create a new spa_t in the namespace. * * spa_remove() Remove a spa_t from the namespace. This also * frees up any memory associated with the spa_t. * * spa_next() Returns the next spa_t in the system, or the * first if NULL is passed. * * spa_evict_all() Shutdown and remove all spa_t structures in * the system. * * spa_guid_exists() Determine whether a pool/device guid exists. * * The spa_refcount is manipulated using the following functions: * * spa_open_ref() Adds a reference to the given spa_t. Must be * called with spa_namespace_lock held if the * refcount is currently zero. * * spa_close() Remove a reference from the spa_t. This will * not free the spa_t or remove it from the * namespace. No locking is required. * * spa_refcount_zero() Returns true if the refcount is currently * zero. Must be called with spa_namespace_lock * held. * * The spa_config_lock[] is an array of rwlocks, ordered as follows: * SCL_CONFIG > SCL_STATE > SCL_ALLOC > SCL_ZIO > SCL_FREE > SCL_VDEV. * spa_config_lock[] is manipulated with spa_config_{enter,exit,held}(). * * To read the configuration, it suffices to hold one of these locks as reader. * To modify the configuration, you must hold all locks as writer. To modify * vdev state without altering the vdev tree's topology (e.g. online/offline), * you must hold SCL_STATE and SCL_ZIO as writer. * * We use these distinct config locks to avoid recursive lock entry. * For example, spa_sync() (which holds SCL_CONFIG as reader) induces * block allocations (SCL_ALLOC), which may require reading space maps * from disk (dmu_read() -> zio_read() -> SCL_ZIO). * * The spa config locks cannot be normal rwlocks because we need the * ability to hand off ownership. For example, SCL_ZIO is acquired * by the issuing thread and later released by an interrupt thread. * They do, however, obey the usual write-wanted semantics to prevent * writer (i.e. system administrator) starvation. * * The lock acquisition rules are as follows: * * SCL_CONFIG * Protects changes to the vdev tree topology, such as vdev * add/remove/attach/detach. Protects the dirty config list * (spa_config_dirty_list) and the set of spares and l2arc devices. * * SCL_STATE * Protects changes to pool state and vdev state, such as vdev * online/offline/fault/degrade/clear. Protects the dirty state list * (spa_state_dirty_list) and global pool state (spa_state). * * SCL_ALLOC * Protects changes to metaslab groups and classes. * Held as reader by metaslab_alloc() and metaslab_claim(). * * SCL_ZIO * Held by bp-level zios (those which have no io_vd upon entry) * to prevent changes to the vdev tree. The bp-level zio implicitly * protects all of its vdev child zios, which do not hold SCL_ZIO. * * SCL_FREE * Protects changes to metaslab groups and classes. * Held as reader by metaslab_free(). SCL_FREE is distinct from * SCL_ALLOC, and lower than SCL_ZIO, so that we can safely free * blocks in zio_done() while another i/o that holds either * SCL_ALLOC or SCL_ZIO is waiting for this i/o to complete. * * SCL_VDEV * Held as reader to prevent changes to the vdev tree during trivial * inquiries such as bp_get_dsize(). SCL_VDEV is distinct from the * other locks, and lower than all of them, to ensure that it's safe * to acquire regardless of caller context. * * In addition, the following rules apply: * * (a) spa_props_lock protects pool properties, spa_config and spa_config_list. * The lock ordering is SCL_CONFIG > spa_props_lock. * * (b) I/O operations on leaf vdevs. For any zio operation that takes * an explicit vdev_t argument -- such as zio_ioctl(), zio_read_phys(), * or zio_write_phys() -- the caller must ensure that the config cannot * cannot change in the interim, and that the vdev cannot be reopened. * SCL_STATE as reader suffices for both. * * The vdev configuration is protected by spa_vdev_enter() / spa_vdev_exit(). * * spa_vdev_enter() Acquire the namespace lock and the config lock * for writing. * * spa_vdev_exit() Release the config lock, wait for all I/O * to complete, sync the updated configs to the * cache, and release the namespace lock. * * vdev state is protected by spa_vdev_state_enter() / spa_vdev_state_exit(). * Like spa_vdev_enter/exit, these are convenience wrappers -- the actual * locking is, always, based on spa_namespace_lock and spa_config_lock[]. */ static avl_tree_t spa_namespace_avl; kmutex_t spa_namespace_lock; static kcondvar_t spa_namespace_cv; static const int spa_max_replication_override = SPA_DVAS_PER_BP; static kmutex_t spa_spare_lock; static avl_tree_t spa_spare_avl; static kmutex_t spa_l2cache_lock; static avl_tree_t spa_l2cache_avl; spa_mode_t spa_mode_global = SPA_MODE_UNINIT; #ifdef ZFS_DEBUG /* * Everything except dprintf, set_error, spa, and indirect_remap is on * by default in debug builds. */ int zfs_flags = ~(ZFS_DEBUG_DPRINTF | ZFS_DEBUG_SET_ERROR | ZFS_DEBUG_INDIRECT_REMAP); #else int zfs_flags = 0; #endif /* * zfs_recover can be set to nonzero to attempt to recover from * otherwise-fatal errors, typically caused by on-disk corruption. When * set, calls to zfs_panic_recover() will turn into warning messages. * This should only be used as a last resort, as it typically results * in leaked space, or worse. */ int zfs_recover = B_FALSE; /* * If destroy encounters an EIO while reading metadata (e.g. indirect * blocks), space referenced by the missing metadata can not be freed. * Normally this causes the background destroy to become "stalled", as * it is unable to make forward progress. While in this stalled state, * all remaining space to free from the error-encountering filesystem is * "temporarily leaked". Set this flag to cause it to ignore the EIO, * permanently leak the space from indirect blocks that can not be read, * and continue to free everything else that it can. * * The default, "stalling" behavior is useful if the storage partially * fails (i.e. some but not all i/os fail), and then later recovers. In * this case, we will be able to continue pool operations while it is * partially failed, and when it recovers, we can continue to free the * space, with no leaks. However, note that this case is actually * fairly rare. * * Typically pools either (a) fail completely (but perhaps temporarily, * e.g. a top-level vdev going offline), or (b) have localized, * permanent errors (e.g. disk returns the wrong data due to bit flip or * firmware bug). In case (a), this setting does not matter because the * pool will be suspended and the sync thread will not be able to make * forward progress regardless. In case (b), because the error is * permanent, the best we can do is leak the minimum amount of space, * which is what setting this flag will do. Therefore, it is reasonable * for this flag to normally be set, but we chose the more conservative * approach of not setting it, so that there is no possibility of * leaking space in the "partial temporary" failure case. */ int zfs_free_leak_on_eio = B_FALSE; /* * Expiration time in milliseconds. This value has two meanings. First it is * used to determine when the spa_deadman() logic should fire. By default the * spa_deadman() will fire if spa_sync() has not completed in 600 seconds. * Secondly, the value determines if an I/O is considered "hung". Any I/O that * has not completed in zfs_deadman_synctime_ms is considered "hung" resulting * in one of three behaviors controlled by zfs_deadman_failmode. */ unsigned long zfs_deadman_synctime_ms = 600000UL; /* 10 min. */ /* * This value controls the maximum amount of time zio_wait() will block for an * outstanding IO. By default this is 300 seconds at which point the "hung" * behavior will be applied as described for zfs_deadman_synctime_ms. */ unsigned long zfs_deadman_ziotime_ms = 300000UL; /* 5 min. */ /* * Check time in milliseconds. This defines the frequency at which we check * for hung I/O. */ unsigned long zfs_deadman_checktime_ms = 60000UL; /* 1 min. */ /* * By default the deadman is enabled. */ int zfs_deadman_enabled = B_TRUE; /* * Controls the behavior of the deadman when it detects a "hung" I/O. * Valid values are zfs_deadman_failmode=. * * wait - Wait for the "hung" I/O (default) * continue - Attempt to recover from a "hung" I/O * panic - Panic the system */ const char *zfs_deadman_failmode = "wait"; /* * The worst case is single-sector max-parity RAID-Z blocks, in which * case the space requirement is exactly (VDEV_RAIDZ_MAXPARITY + 1) * times the size; so just assume that. Add to this the fact that * we can have up to 3 DVAs per bp, and one more factor of 2 because * the block may be dittoed with up to 3 DVAs by ddt_sync(). All together, * the worst case is: * (VDEV_RAIDZ_MAXPARITY + 1) * SPA_DVAS_PER_BP * 2 == 24 */ uint_t spa_asize_inflation = 24; /* * Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space in * the pool to be consumed (bounded by spa_max_slop). This ensures that we * don't run the pool completely out of space, due to unaccounted changes (e.g. * to the MOS). It also limits the worst-case time to allocate space. If we * have less than this amount of free space, most ZPL operations (e.g. write, * create) will return ENOSPC. The ZIL metaslabs (spa_embedded_log_class) are * also part of this 3.2% of space which can't be consumed by normal writes; * the slop space "proper" (spa_get_slop_space()) is decreased by the embedded * log space. * * Certain operations (e.g. file removal, most administrative actions) can * use half the slop space. They will only return ENOSPC if less than half * the slop space is free. Typically, once the pool has less than the slop * space free, the user will use these operations to free up space in the pool. * These are the operations that call dsl_pool_adjustedsize() with the netfree * argument set to TRUE. * * Operations that are almost guaranteed to free up space in the absence of * a pool checkpoint can use up to three quarters of the slop space * (e.g zfs destroy). * * A very restricted set of operations are always permitted, regardless of * the amount of free space. These are the operations that call * dsl_sync_task(ZFS_SPACE_CHECK_NONE). If these operations result in a net * increase in the amount of space used, it is possible to run the pool * completely out of space, causing it to be permanently read-only. * * Note that on very small pools, the slop space will be larger than * 3.2%, in an effort to have it be at least spa_min_slop (128MB), * but we never allow it to be more than half the pool size. * * Further, on very large pools, the slop space will be smaller than * 3.2%, to avoid reserving much more space than we actually need; bounded * by spa_max_slop (128GB). * * See also the comments in zfs_space_check_t. */ uint_t spa_slop_shift = 5; static const uint64_t spa_min_slop = 128ULL * 1024 * 1024; static const uint64_t spa_max_slop = 128ULL * 1024 * 1024 * 1024; static const int spa_allocators = 4; void spa_load_failed(spa_t *spa, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); zfs_dbgmsg("spa_load(%s, config %s): FAILED: %s", spa->spa_name, spa->spa_trust_config ? "trusted" : "untrusted", buf); } void spa_load_note(spa_t *spa, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); zfs_dbgmsg("spa_load(%s, config %s): %s", spa->spa_name, spa->spa_trust_config ? "trusted" : "untrusted", buf); } /* * By default dedup and user data indirects land in the special class */ static int zfs_ddt_data_is_special = B_TRUE; static int zfs_user_indirect_is_special = B_TRUE; /* * The percentage of special class final space reserved for metadata only. * Once we allocate 100 - zfs_special_class_metadata_reserve_pct we only * let metadata into the class. */ static uint_t zfs_special_class_metadata_reserve_pct = 25; /* * ========================================================================== * SPA config locking * ========================================================================== */ static void spa_config_lock_init(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_init(&scl->scl_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&scl->scl_cv, NULL, CV_DEFAULT, NULL); scl->scl_writer = NULL; scl->scl_write_wanted = 0; scl->scl_count = 0; } } static void spa_config_lock_destroy(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_destroy(&scl->scl_lock); cv_destroy(&scl->scl_cv); ASSERT(scl->scl_writer == NULL); ASSERT(scl->scl_write_wanted == 0); ASSERT(scl->scl_count == 0); } } int spa_config_tryenter(spa_t *spa, int locks, const void *tag, krw_t rw) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { if (scl->scl_writer || scl->scl_write_wanted) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } } else { ASSERT(scl->scl_writer != curthread); if (scl->scl_count != 0) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } scl->scl_writer = curthread; } scl->scl_count++; mutex_exit(&scl->scl_lock); } return (1); } void spa_config_enter(spa_t *spa, int locks, const void *tag, krw_t rw) { (void) tag; int wlocks_held = 0; ASSERT3U(SCL_LOCKS, <, sizeof (wlocks_held) * NBBY); for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (scl->scl_writer == curthread) wlocks_held |= (1 << i); if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { while (scl->scl_writer || scl->scl_write_wanted) { cv_wait(&scl->scl_cv, &scl->scl_lock); } } else { ASSERT(scl->scl_writer != curthread); while (scl->scl_count != 0) { scl->scl_write_wanted++; cv_wait(&scl->scl_cv, &scl->scl_lock); scl->scl_write_wanted--; } scl->scl_writer = curthread; } scl->scl_count++; mutex_exit(&scl->scl_lock); } ASSERT3U(wlocks_held, <=, locks); } void spa_config_exit(spa_t *spa, int locks, const void *tag) { (void) tag; for (int i = SCL_LOCKS - 1; i >= 0; i--) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); ASSERT(scl->scl_count > 0); if (--scl->scl_count == 0) { ASSERT(scl->scl_writer == NULL || scl->scl_writer == curthread); scl->scl_writer = NULL; /* OK in either case */ cv_broadcast(&scl->scl_cv); } mutex_exit(&scl->scl_lock); } } int spa_config_held(spa_t *spa, int locks, krw_t rw) { int locks_held = 0; for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; if ((rw == RW_READER && scl->scl_count != 0) || (rw == RW_WRITER && scl->scl_writer == curthread)) locks_held |= 1 << i; } return (locks_held); } /* * ========================================================================== * SPA namespace functions * ========================================================================== */ /* * Lookup the named spa_t in the AVL tree. The spa_namespace_lock must be held. * Returns NULL if no matching spa_t is found. */ spa_t * spa_lookup(const char *name) { static spa_t search; /* spa_t is large; don't allocate on stack */ spa_t *spa; avl_index_t where; char *cp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); (void) strlcpy(search.spa_name, name, sizeof (search.spa_name)); /* * If it's a full dataset name, figure out the pool name and * just use that. */ cp = strpbrk(search.spa_name, "/@#"); if (cp != NULL) *cp = '\0'; spa = avl_find(&spa_namespace_avl, &search, &where); return (spa); } /* * Fires when spa_sync has not completed within zfs_deadman_synctime_ms. * If the zfs_deadman_enabled flag is set then it inspects all vdev queues * looking for potentially hung I/Os. */ void spa_deadman(void *arg) { spa_t *spa = arg; /* Disable the deadman if the pool is suspended. */ if (spa_suspended(spa)) return; zfs_dbgmsg("slow spa_sync: started %llu seconds ago, calls %llu", (gethrtime() - spa->spa_sync_starttime) / NANOSEC, (u_longlong_t)++spa->spa_deadman_calls); if (zfs_deadman_enabled) vdev_deadman(spa->spa_root_vdev, FTAG); spa->spa_deadman_tqid = taskq_dispatch_delay(system_delay_taskq, spa_deadman, spa, TQ_SLEEP, ddi_get_lbolt() + MSEC_TO_TICK(zfs_deadman_checktime_ms)); } static int spa_log_sm_sort_by_txg(const void *va, const void *vb) { const spa_log_sm_t *a = va; const spa_log_sm_t *b = vb; return (TREE_CMP(a->sls_txg, b->sls_txg)); } /* * Create an uninitialized spa_t with the given name. Requires * spa_namespace_lock. The caller must ensure that the spa_t doesn't already * exist by calling spa_lookup() first. */ spa_t * spa_add(const char *name, nvlist_t *config, const char *altroot) { spa_t *spa; spa_config_dirent_t *dp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa = kmem_zalloc(sizeof (spa_t), KM_SLEEP); mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_evicting_os_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_history_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_proc_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_cksum_tmpls_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_suspend_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_vdev_top_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_feat_stats_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_flushed_ms_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_activities_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_evicting_os_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_proc_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_activities_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_waiters_cv, NULL, CV_DEFAULT, NULL); for (int t = 0; t < TXG_SIZE; t++) bplist_create(&spa->spa_free_bplist[t]); (void) strlcpy(spa->spa_name, name, sizeof (spa->spa_name)); spa->spa_state = POOL_STATE_UNINITIALIZED; spa->spa_freeze_txg = UINT64_MAX; spa->spa_final_txg = UINT64_MAX; spa->spa_load_max_txg = UINT64_MAX; spa->spa_proc = &p0; spa->spa_proc_state = SPA_PROC_NONE; spa->spa_trust_config = B_TRUE; spa->spa_hostid = zone_get_hostid(NULL); spa->spa_deadman_synctime = MSEC2NSEC(zfs_deadman_synctime_ms); spa->spa_deadman_ziotime = MSEC2NSEC(zfs_deadman_ziotime_ms); spa_set_deadman_failmode(spa, zfs_deadman_failmode); zfs_refcount_create(&spa->spa_refcount); spa_config_lock_init(spa); spa_stats_init(spa); avl_add(&spa_namespace_avl, spa); /* * Set the alternate root, if there is one. */ if (altroot) spa->spa_root = spa_strdup(altroot); spa->spa_alloc_count = spa_allocators; spa->spa_allocs = kmem_zalloc(spa->spa_alloc_count * sizeof (spa_alloc_t), KM_SLEEP); for (int i = 0; i < spa->spa_alloc_count; i++) { mutex_init(&spa->spa_allocs[i].spaa_lock, NULL, MUTEX_DEFAULT, NULL); avl_create(&spa->spa_allocs[i].spaa_tree, zio_bookmark_compare, sizeof (zio_t), offsetof(zio_t, io_alloc_node)); } avl_create(&spa->spa_metaslabs_by_flushed, metaslab_sort_by_flushed, sizeof (metaslab_t), offsetof(metaslab_t, ms_spa_txg_node)); avl_create(&spa->spa_sm_logs_by_txg, spa_log_sm_sort_by_txg, sizeof (spa_log_sm_t), offsetof(spa_log_sm_t, sls_node)); list_create(&spa->spa_log_summary, sizeof (log_summary_entry_t), offsetof(log_summary_entry_t, lse_node)); /* * Every pool starts with the default cachefile */ list_create(&spa->spa_config_list, sizeof (spa_config_dirent_t), offsetof(spa_config_dirent_t, scd_link)); dp = kmem_zalloc(sizeof (spa_config_dirent_t), KM_SLEEP); dp->scd_path = altroot ? NULL : spa_strdup(spa_config_path); list_insert_head(&spa->spa_config_list, dp); VERIFY(nvlist_alloc(&spa->spa_load_info, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (config != NULL) { nvlist_t *features; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) == 0) { VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } VERIFY(nvlist_dup(config, &spa->spa_config, 0) == 0); } if (spa->spa_label_features == NULL) { VERIFY(nvlist_alloc(&spa->spa_label_features, NV_UNIQUE_NAME, KM_SLEEP) == 0); } spa->spa_min_ashift = INT_MAX; spa->spa_max_ashift = 0; spa->spa_min_alloc = INT_MAX; /* Reset cached value */ spa->spa_dedup_dspace = ~0ULL; /* * As a pool is being created, treat all features as disabled by * setting SPA_FEATURE_DISABLED for all entries in the feature * refcount cache. */ for (int i = 0; i < SPA_FEATURES; i++) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } list_create(&spa->spa_leaf_list, sizeof (vdev_t), offsetof(vdev_t, vdev_leaf_node)); return (spa); } /* * Removes a spa_t from the namespace, freeing up any memory used. Requires * spa_namespace_lock. This is called only after the spa_t has been closed and * deactivated. */ void spa_remove(spa_t *spa) { spa_config_dirent_t *dp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_state(spa) == POOL_STATE_UNINITIALIZED); ASSERT3U(zfs_refcount_count(&spa->spa_refcount), ==, 0); ASSERT0(spa->spa_waiters); nvlist_free(spa->spa_config_splitting); avl_remove(&spa_namespace_avl, spa); cv_broadcast(&spa_namespace_cv); if (spa->spa_root) spa_strfree(spa->spa_root); while ((dp = list_head(&spa->spa_config_list)) != NULL) { list_remove(&spa->spa_config_list, dp); if (dp->scd_path != NULL) spa_strfree(dp->scd_path); kmem_free(dp, sizeof (spa_config_dirent_t)); } for (int i = 0; i < spa->spa_alloc_count; i++) { avl_destroy(&spa->spa_allocs[i].spaa_tree); mutex_destroy(&spa->spa_allocs[i].spaa_lock); } kmem_free(spa->spa_allocs, spa->spa_alloc_count * sizeof (spa_alloc_t)); avl_destroy(&spa->spa_metaslabs_by_flushed); avl_destroy(&spa->spa_sm_logs_by_txg); list_destroy(&spa->spa_log_summary); list_destroy(&spa->spa_config_list); list_destroy(&spa->spa_leaf_list); nvlist_free(spa->spa_label_features); nvlist_free(spa->spa_load_info); nvlist_free(spa->spa_feat_stats); spa_config_set(spa, NULL); zfs_refcount_destroy(&spa->spa_refcount); spa_stats_destroy(spa); spa_config_lock_destroy(spa); for (int t = 0; t < TXG_SIZE; t++) bplist_destroy(&spa->spa_free_bplist[t]); zio_checksum_templates_free(spa); cv_destroy(&spa->spa_async_cv); cv_destroy(&spa->spa_evicting_os_cv); cv_destroy(&spa->spa_proc_cv); cv_destroy(&spa->spa_scrub_io_cv); cv_destroy(&spa->spa_suspend_cv); cv_destroy(&spa->spa_activities_cv); cv_destroy(&spa->spa_waiters_cv); mutex_destroy(&spa->spa_flushed_ms_lock); mutex_destroy(&spa->spa_async_lock); mutex_destroy(&spa->spa_errlist_lock); mutex_destroy(&spa->spa_errlog_lock); mutex_destroy(&spa->spa_evicting_os_lock); mutex_destroy(&spa->spa_history_lock); mutex_destroy(&spa->spa_proc_lock); mutex_destroy(&spa->spa_props_lock); mutex_destroy(&spa->spa_cksum_tmpls_lock); mutex_destroy(&spa->spa_scrub_lock); mutex_destroy(&spa->spa_suspend_lock); mutex_destroy(&spa->spa_vdev_top_lock); mutex_destroy(&spa->spa_feat_stats_lock); mutex_destroy(&spa->spa_activities_lock); kmem_free(spa, sizeof (spa_t)); } /* * Given a pool, return the next pool in the namespace, or NULL if there is * none. If 'prev' is NULL, return the first pool. */ spa_t * spa_next(spa_t *prev) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (prev) return (AVL_NEXT(&spa_namespace_avl, prev)); else return (avl_first(&spa_namespace_avl)); } /* * ========================================================================== * SPA refcount functions * ========================================================================== */ /* * Add a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_open_ref(spa_t *spa, const void *tag) { ASSERT(zfs_refcount_count(&spa->spa_refcount) >= spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) zfs_refcount_add(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_close(spa_t *spa, const void *tag) { ASSERT(zfs_refcount_count(&spa->spa_refcount) > spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) zfs_refcount_remove(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t held by a dsl dir that is * being asynchronously released. Async releases occur from a taskq * performing eviction of dsl datasets and dirs. The namespace lock * isn't held and the hold by the object being evicted may contribute to * spa_minref (e.g. dataset or directory released during pool export), * so the asserts in spa_close() do not apply. */ void spa_async_close(spa_t *spa, const void *tag) { (void) zfs_refcount_remove(&spa->spa_refcount, tag); } /* * Check to see if the spa refcount is zero. Must be called with * spa_namespace_lock held. We really compare against spa_minref, which is the * number of references acquired when opening a pool */ boolean_t spa_refcount_zero(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); return (zfs_refcount_count(&spa->spa_refcount) == spa->spa_minref); } /* * ========================================================================== * SPA spare and l2cache tracking * ========================================================================== */ /* * Hot spares and cache devices are tracked using the same code below, * for 'auxiliary' devices. */ typedef struct spa_aux { uint64_t aux_guid; uint64_t aux_pool; avl_node_t aux_avl; int aux_count; } spa_aux_t; static inline int spa_aux_compare(const void *a, const void *b) { const spa_aux_t *sa = (const spa_aux_t *)a; const spa_aux_t *sb = (const spa_aux_t *)b; return (TREE_CMP(sa->aux_guid, sb->aux_guid)); } static void spa_aux_add(vdev_t *vd, avl_tree_t *avl) { avl_index_t where; spa_aux_t search; spa_aux_t *aux; search.aux_guid = vd->vdev_guid; if ((aux = avl_find(avl, &search, &where)) != NULL) { aux->aux_count++; } else { aux = kmem_zalloc(sizeof (spa_aux_t), KM_SLEEP); aux->aux_guid = vd->vdev_guid; aux->aux_count = 1; avl_insert(avl, aux, where); } } static void spa_aux_remove(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search; spa_aux_t *aux; avl_index_t where; search.aux_guid = vd->vdev_guid; aux = avl_find(avl, &search, &where); ASSERT(aux != NULL); if (--aux->aux_count == 0) { avl_remove(avl, aux); kmem_free(aux, sizeof (spa_aux_t)); } else if (aux->aux_pool == spa_guid(vd->vdev_spa)) { aux->aux_pool = 0ULL; } } static boolean_t spa_aux_exists(uint64_t guid, uint64_t *pool, int *refcnt, avl_tree_t *avl) { spa_aux_t search, *found; search.aux_guid = guid; found = avl_find(avl, &search, NULL); if (pool) { if (found) *pool = found->aux_pool; else *pool = 0ULL; } if (refcnt) { if (found) *refcnt = found->aux_count; else *refcnt = 0; } return (found != NULL); } static void spa_aux_activate(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search, *found; avl_index_t where; search.aux_guid = vd->vdev_guid; found = avl_find(avl, &search, &where); ASSERT(found != NULL); ASSERT(found->aux_pool == 0ULL); found->aux_pool = spa_guid(vd->vdev_spa); } /* * Spares are tracked globally due to the following constraints: * * - A spare may be part of multiple pools. * - A spare may be added to a pool even if it's actively in use within * another pool. * - A spare in use in any pool can only be the source of a replacement if * the target is a spare in the same pool. * * We keep track of all spares on the system through the use of a reference * counted AVL tree. When a vdev is added as a spare, or used as a replacement * spare, then we bump the reference count in the AVL tree. In addition, we set * the 'vdev_isspare' member to indicate that the device is a spare (active or * inactive). When a spare is made active (used to replace a device in the * pool), we also keep track of which pool its been made a part of. * * The 'spa_spare_lock' protects the AVL tree. These functions are normally * called under the spa_namespace lock as part of vdev reconfiguration. The * separate spare lock exists for the status query path, which does not need to * be completely consistent with respect to other vdev configuration changes. */ static int spa_spare_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_spare_add(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(!vd->vdev_isspare); spa_aux_add(vd, &spa_spare_avl); vd->vdev_isspare = B_TRUE; mutex_exit(&spa_spare_lock); } void spa_spare_remove(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_remove(vd, &spa_spare_avl); vd->vdev_isspare = B_FALSE; mutex_exit(&spa_spare_lock); } boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt) { boolean_t found; mutex_enter(&spa_spare_lock); found = spa_aux_exists(guid, pool, refcnt, &spa_spare_avl); mutex_exit(&spa_spare_lock); return (found); } void spa_spare_activate(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_activate(vd, &spa_spare_avl); mutex_exit(&spa_spare_lock); } /* * Level 2 ARC devices are tracked globally for the same reasons as spares. * Cache devices currently only support one pool per cache device, and so * for these devices the aux reference count is currently unused beyond 1. */ static int spa_l2cache_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_l2cache_add(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(!vd->vdev_isl2cache); spa_aux_add(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_TRUE; mutex_exit(&spa_l2cache_lock); } void spa_l2cache_remove(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_remove(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_FALSE; mutex_exit(&spa_l2cache_lock); } boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool) { boolean_t found; mutex_enter(&spa_l2cache_lock); found = spa_aux_exists(guid, pool, NULL, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); return (found); } void spa_l2cache_activate(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_activate(vd, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); } /* * ========================================================================== * SPA vdev locking * ========================================================================== */ /* * Lock the given spa_t for the purpose of adding or removing a vdev. * Grabs the global spa_namespace_lock plus the spa config lock for writing. * It returns the next transaction group for the spa_t. */ uint64_t spa_vdev_enter(spa_t *spa) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); vdev_autotrim_stop_all(spa); return (spa_vdev_config_enter(spa)); } /* * The same as spa_vdev_enter() above but additionally takes the guid of * the vdev being detached. When there is a rebuild in process it will be * suspended while the vdev tree is modified then resumed by spa_vdev_exit(). * The rebuild is canceled if only a single child remains after the detach. */ uint64_t spa_vdev_detach_enter(spa_t *spa, uint64_t guid) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); vdev_autotrim_stop_all(spa); if (guid != 0) { vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd) { vdev_rebuild_stop_wait(vd->vdev_top); } } return (spa_vdev_config_enter(spa)); } /* * Internal implementation for spa_vdev_enter(). Used when a vdev * operation requires multiple syncs (i.e. removing a device) while * keeping the spa_namespace_lock held. */ uint64_t spa_vdev_config_enter(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); return (spa_last_synced_txg(spa) + 1); } /* * Used in combination with spa_vdev_config_enter() to allow the syncing * of multiple transactions without releasing the spa_namespace_lock. */ void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, const char *tag) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); int config_changed = B_FALSE; ASSERT(txg > spa_last_synced_txg(spa)); spa->spa_pending_vdev = NULL; /* * Reassess the DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, 0, 0, B_FALSE, B_FALSE); if (error == 0 && !list_is_empty(&spa->spa_config_dirty_list)) { config_changed = B_TRUE; spa->spa_config_generation++; } /* * Verify the metaslab classes. */ ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_embedded_log_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_special_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_dedup_class(spa)) == 0); spa_config_exit(spa, SCL_ALL, spa); /* * Panic the system if the specified tag requires it. This * is useful for ensuring that configurations are updated * transactionally. */ if (zio_injection_enabled) zio_handle_panic_injection(spa, tag, 0); /* * Note: this txg_wait_synced() is important because it ensures * that there won't be more than one config change per txg. * This allows us to use the txg as the generation number. */ if (error == 0) txg_wait_synced(spa->spa_dsl_pool, txg); if (vd != NULL) { ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL); if (vd->vdev_ops->vdev_op_leaf) { mutex_enter(&vd->vdev_initialize_lock); vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED, NULL); mutex_exit(&vd->vdev_initialize_lock); mutex_enter(&vd->vdev_trim_lock); vdev_trim_stop(vd, VDEV_TRIM_CANCELED, NULL); mutex_exit(&vd->vdev_trim_lock); } /* * The vdev may be both a leaf and top-level device. */ vdev_autotrim_stop_wait(vd); spa_config_enter(spa, SCL_STATE_ALL, spa, RW_WRITER); vdev_free(vd); spa_config_exit(spa, SCL_STATE_ALL, spa); } /* * If the config changed, update the config cache. */ if (config_changed) - spa_write_cachefile(spa, B_FALSE, B_TRUE); + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_FALSE); } /* * Unlock the spa_t after adding or removing a vdev. Besides undoing the * locking of spa_vdev_enter(), we also want make sure the transactions have * synced to disk, and then update the global configuration cache with the new * information. */ int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error) { vdev_autotrim_restart(spa); vdev_rebuild_restart(spa); spa_vdev_config_exit(spa, vd, txg, error, FTAG); mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Lock the given spa_t for the purpose of changing vdev state. */ void spa_vdev_state_enter(spa_t *spa, int oplocks) { int locks = SCL_STATE_ALL | oplocks; /* * Root pools may need to read of the underlying devfs filesystem * when opening up a vdev. Unfortunately if we're holding the * SCL_ZIO lock it will result in a deadlock when we try to issue * the read from the root filesystem. Instead we "prefetch" * the associated vnodes that we need prior to opening the * underlying devices and cache them so that we can prevent * any I/O when we are doing the actual open. */ if (spa_is_root(spa)) { int low = locks & ~(SCL_ZIO - 1); int high = locks & ~low; spa_config_enter(spa, high, spa, RW_WRITER); vdev_hold(spa->spa_root_vdev); spa_config_enter(spa, low, spa, RW_WRITER); } else { spa_config_enter(spa, locks, spa, RW_WRITER); } spa->spa_vdev_locks = locks; } int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error) { boolean_t config_changed = B_FALSE; vdev_t *vdev_top; if (vd == NULL || vd == spa->spa_root_vdev) { vdev_top = spa->spa_root_vdev; } else { vdev_top = vd->vdev_top; } if (vd != NULL || error == 0) vdev_dtl_reassess(vdev_top, 0, 0, B_FALSE, B_FALSE); if (vd != NULL) { if (vd != spa->spa_root_vdev) vdev_state_dirty(vdev_top); config_changed = B_TRUE; spa->spa_config_generation++; } if (spa_is_root(spa)) vdev_rele(spa->spa_root_vdev); ASSERT3U(spa->spa_vdev_locks, >=, SCL_STATE_ALL); spa_config_exit(spa, spa->spa_vdev_locks, spa); /* * If anything changed, wait for it to sync. This ensures that, * from the system administrator's perspective, zpool(8) commands * are synchronous. This is important for things like zpool offline: * when the command completes, you expect no further I/O from ZFS. */ if (vd != NULL) txg_wait_synced(spa->spa_dsl_pool, 0); /* * If the config changed, update the config cache. */ if (config_changed) { mutex_enter(&spa_namespace_lock); - spa_write_cachefile(spa, B_FALSE, B_TRUE); + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_FALSE); mutex_exit(&spa_namespace_lock); } return (error); } /* * ========================================================================== * Miscellaneous functions * ========================================================================== */ void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx) { if (!nvlist_exists(spa->spa_label_features, feature)) { fnvlist_add_boolean(spa->spa_label_features, feature); /* * When we are creating the pool (tx_txg==TXG_INITIAL), we can't * dirty the vdev config because lock SCL_CONFIG is not held. * Thankfully, in this case we don't need to dirty the config * because it will be written out anyway when we finish * creating the pool. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); } } void spa_deactivate_mos_feature(spa_t *spa, const char *feature) { if (nvlist_remove_all(spa->spa_label_features, feature) == 0) vdev_config_dirty(spa->spa_root_vdev); } /* * Return the spa_t associated with given pool_guid, if it exists. If * device_guid is non-zero, determine whether the pool exists *and* contains * a device with the specified device_guid. */ spa_t * spa_by_guid(uint64_t pool_guid, uint64_t device_guid) { spa_t *spa; avl_tree_t *t = &spa_namespace_avl; ASSERT(MUTEX_HELD(&spa_namespace_lock)); for (spa = avl_first(t); spa != NULL; spa = AVL_NEXT(t, spa)) { if (spa->spa_state == POOL_STATE_UNINITIALIZED) continue; if (spa->spa_root_vdev == NULL) continue; if (spa_guid(spa) == pool_guid) { if (device_guid == 0) break; if (vdev_lookup_by_guid(spa->spa_root_vdev, device_guid) != NULL) break; /* * Check any devices we may be in the process of adding. */ if (spa->spa_pending_vdev) { if (vdev_lookup_by_guid(spa->spa_pending_vdev, device_guid) != NULL) break; } } } return (spa); } /* * Determine whether a pool with the given pool_guid exists. */ boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid) { return (spa_by_guid(pool_guid, device_guid) != NULL); } char * spa_strdup(const char *s) { size_t len; char *new; len = strlen(s); new = kmem_alloc(len + 1, KM_SLEEP); memcpy(new, s, len + 1); return (new); } void spa_strfree(char *s) { kmem_free(s, strlen(s) + 1); } uint64_t spa_generate_guid(spa_t *spa) { uint64_t guid; if (spa != NULL) { do { (void) random_get_pseudo_bytes((void *)&guid, sizeof (guid)); } while (guid == 0 || spa_guid_exists(spa_guid(spa), guid)); } else { do { (void) random_get_pseudo_bytes((void *)&guid, sizeof (guid)); } while (guid == 0 || spa_guid_exists(guid, 0)); } return (guid); } void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp) { char type[256]; const char *checksum = NULL; const char *compress = NULL; if (bp != NULL) { if (BP_GET_TYPE(bp) & DMU_OT_NEWTYPE) { dmu_object_byteswap_t bswap = DMU_OT_BYTESWAP(BP_GET_TYPE(bp)); (void) snprintf(type, sizeof (type), "bswap %s %s", DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) ? "metadata" : "data", dmu_ot_byteswap[bswap].ob_name); } else { (void) strlcpy(type, dmu_ot[BP_GET_TYPE(bp)].ot_name, sizeof (type)); } if (!BP_IS_EMBEDDED(bp)) { checksum = zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name; } compress = zio_compress_table[BP_GET_COMPRESS(bp)].ci_name; } SNPRINTF_BLKPTR(snprintf, ' ', buf, buflen, bp, type, checksum, compress); } void spa_freeze(spa_t *spa) { uint64_t freeze_txg = 0; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); if (spa->spa_freeze_txg == UINT64_MAX) { freeze_txg = spa_last_synced_txg(spa) + TXG_SIZE; spa->spa_freeze_txg = freeze_txg; } spa_config_exit(spa, SCL_ALL, FTAG); if (freeze_txg != 0) txg_wait_synced(spa_get_dsl(spa), freeze_txg); } void zfs_panic_recover(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx); va_end(adx); } /* * This is a stripped-down version of strtoull, suitable only for converting * lowercase hexadecimal numbers that don't overflow. */ uint64_t zfs_strtonum(const char *str, char **nptr) { uint64_t val = 0; char c; int digit; while ((c = *str) != '\0') { if (c >= '0' && c <= '9') digit = c - '0'; else if (c >= 'a' && c <= 'f') digit = 10 + c - 'a'; else break; val *= 16; val += digit; str++; } if (nptr) *nptr = (char *)str; return (val); } void spa_activate_allocation_classes(spa_t *spa, dmu_tx_t *tx) { /* * We bump the feature refcount for each special vdev added to the pool */ ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)); spa_feature_incr(spa, SPA_FEATURE_ALLOCATION_CLASSES, tx); } /* * ========================================================================== * Accessor functions * ========================================================================== */ boolean_t spa_shutting_down(spa_t *spa) { return (spa->spa_async_suspended); } dsl_pool_t * spa_get_dsl(spa_t *spa) { return (spa->spa_dsl_pool); } boolean_t spa_is_initializing(spa_t *spa) { return (spa->spa_is_initializing); } boolean_t spa_indirect_vdevs_loaded(spa_t *spa) { return (spa->spa_indirect_vdevs_loaded); } blkptr_t * spa_get_rootblkptr(spa_t *spa) { return (&spa->spa_ubsync.ub_rootbp); } void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp) { spa->spa_uberblock.ub_rootbp = *bp; } void spa_altroot(spa_t *spa, char *buf, size_t buflen) { if (spa->spa_root == NULL) buf[0] = '\0'; else (void) strlcpy(buf, spa->spa_root, buflen); } uint32_t spa_sync_pass(spa_t *spa) { return (spa->spa_sync_pass); } char * spa_name(spa_t *spa) { return (spa->spa_name); } uint64_t spa_guid(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t guid; /* * If we fail to parse the config during spa_load(), we can go through * the error path (which posts an ereport) and end up here with no root * vdev. We stash the original pool guid in 'spa_config_guid' to handle * this case. */ if (spa->spa_root_vdev == NULL) return (spa->spa_config_guid); guid = spa->spa_last_synced_guid != 0 ? spa->spa_last_synced_guid : spa->spa_root_vdev->vdev_guid; /* * Return the most recently synced out guid unless we're * in syncing context. */ if (dp && dsl_pool_sync_context(dp)) return (spa->spa_root_vdev->vdev_guid); else return (guid); } uint64_t spa_load_guid(spa_t *spa) { /* * This is a GUID that exists solely as a reference for the * purposes of the arc. It is generated at load time, and * is never written to persistent storage. */ return (spa->spa_load_guid); } uint64_t spa_last_synced_txg(spa_t *spa) { return (spa->spa_ubsync.ub_txg); } uint64_t spa_first_txg(spa_t *spa) { return (spa->spa_first_txg); } uint64_t spa_syncing_txg(spa_t *spa) { return (spa->spa_syncing_txg); } /* * Return the last txg where data can be dirtied. The final txgs * will be used to just clear out any deferred frees that remain. */ uint64_t spa_final_dirty_txg(spa_t *spa) { return (spa->spa_final_txg - TXG_DEFER_SIZE); } pool_state_t spa_state(spa_t *spa) { return (spa->spa_state); } spa_load_state_t spa_load_state(spa_t *spa) { return (spa->spa_load_state); } uint64_t spa_freeze_txg(spa_t *spa) { return (spa->spa_freeze_txg); } /* * Return the inflated asize for a logical write in bytes. This is used by the * DMU to calculate the space a logical write will require on disk. * If lsize is smaller than the largest physical block size allocatable on this * pool we use its value instead, since the write will end up using the whole * block anyway. */ uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize) { if (lsize == 0) return (0); /* No inflation needed */ return (MAX(lsize, 1 << spa->spa_max_ashift) * spa_asize_inflation); } /* * Return the amount of slop space in bytes. It is typically 1/32 of the pool * (3.2%), minus the embedded log space. On very small pools, it may be * slightly larger than this. On very large pools, it will be capped to * the value of spa_max_slop. The embedded log space is not included in * spa_dspace. By subtracting it, the usable space (per "zfs list") is a * constant 97% of the total space, regardless of metaslab size (assuming the * default spa_slop_shift=5 and a non-tiny pool). * * See the comment above spa_slop_shift for more details. */ uint64_t spa_get_slop_space(spa_t *spa) { uint64_t space = 0; uint64_t slop = 0; /* * Make sure spa_dedup_dspace has been set. */ if (spa->spa_dedup_dspace == ~0ULL) spa_update_dspace(spa); /* * spa_get_dspace() includes the space only logically "used" by * deduplicated data, so since it's not useful to reserve more * space with more deduplicated data, we subtract that out here. */ space = spa_get_dspace(spa) - spa->spa_dedup_dspace; slop = MIN(space >> spa_slop_shift, spa_max_slop); /* * Subtract the embedded log space, but no more than half the (3.2%) * unusable space. Note, the "no more than half" is only relevant if * zfs_embedded_slog_min_ms >> spa_slop_shift < 2, which is not true by * default. */ uint64_t embedded_log = metaslab_class_get_dspace(spa_embedded_log_class(spa)); slop -= MIN(embedded_log, slop >> 1); /* * Slop space should be at least spa_min_slop, but no more than half * the entire pool. */ slop = MAX(slop, MIN(space >> 1, spa_min_slop)); return (slop); } uint64_t spa_get_dspace(spa_t *spa) { return (spa->spa_dspace); } uint64_t spa_get_checkpoint_space(spa_t *spa) { return (spa->spa_checkpoint_info.sci_dspace); } void spa_update_dspace(spa_t *spa) { spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) + ddt_get_dedup_dspace(spa); if (spa->spa_nonallocating_dspace > 0) { /* * Subtract the space provided by all non-allocating vdevs that * contribute to dspace. If a file is overwritten, its old * blocks are freed and new blocks are allocated. If there are * no snapshots of the file, the available space should remain * the same. The old blocks could be freed from the * non-allocating vdev, but the new blocks must be allocated on * other (allocating) vdevs. By reserving the entire size of * the non-allocating vdevs (including allocated space), we * ensure that there will be enough space on the allocating * vdevs for this file overwrite to succeed. * * Note that the DMU/DSL doesn't actually know or care * how much space is allocated (it does its own tracking * of how much space has been logically used). So it * doesn't matter that the data we are moving may be * allocated twice (on the old device and the new device). */ ASSERT3U(spa->spa_dspace, >=, spa->spa_nonallocating_dspace); spa->spa_dspace -= spa->spa_nonallocating_dspace; } } /* * Return the failure mode that has been set to this pool. The default * behavior will be to block all I/Os when a complete failure occurs. */ uint64_t spa_get_failmode(spa_t *spa) { return (spa->spa_failmode); } boolean_t spa_suspended(spa_t *spa) { return (spa->spa_suspended != ZIO_SUSPEND_NONE); } uint64_t spa_version(spa_t *spa) { return (spa->spa_ubsync.ub_version); } boolean_t spa_deflate(spa_t *spa) { return (spa->spa_deflate); } metaslab_class_t * spa_normal_class(spa_t *spa) { return (spa->spa_normal_class); } metaslab_class_t * spa_log_class(spa_t *spa) { return (spa->spa_log_class); } metaslab_class_t * spa_embedded_log_class(spa_t *spa) { return (spa->spa_embedded_log_class); } metaslab_class_t * spa_special_class(spa_t *spa) { return (spa->spa_special_class); } metaslab_class_t * spa_dedup_class(spa_t *spa) { return (spa->spa_dedup_class); } /* * Locate an appropriate allocation class */ metaslab_class_t * spa_preferred_class(spa_t *spa, uint64_t size, dmu_object_type_t objtype, uint_t level, uint_t special_smallblk) { /* * ZIL allocations determine their class in zio_alloc_zil(). */ ASSERT(objtype != DMU_OT_INTENT_LOG); boolean_t has_special_class = spa->spa_special_class->mc_groups != 0; if (DMU_OT_IS_DDT(objtype)) { if (spa->spa_dedup_class->mc_groups != 0) return (spa_dedup_class(spa)); else if (has_special_class && zfs_ddt_data_is_special) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } /* Indirect blocks for user data can land in special if allowed */ if (level > 0 && (DMU_OT_IS_FILE(objtype) || objtype == DMU_OT_ZVOL)) { if (has_special_class && zfs_user_indirect_is_special) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } if (DMU_OT_IS_METADATA(objtype) || level > 0) { if (has_special_class) return (spa_special_class(spa)); else return (spa_normal_class(spa)); } /* * Allow small file blocks in special class in some cases (like * for the dRAID vdev feature). But always leave a reserve of * zfs_special_class_metadata_reserve_pct exclusively for metadata. */ if (DMU_OT_IS_FILE(objtype) && has_special_class && size <= special_smallblk) { metaslab_class_t *special = spa_special_class(spa); uint64_t alloc = metaslab_class_get_alloc(special); uint64_t space = metaslab_class_get_space(special); uint64_t limit = (space * (100 - zfs_special_class_metadata_reserve_pct)) / 100; if (alloc < limit) return (special); } return (spa_normal_class(spa)); } void spa_evicting_os_register(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_insert_head(&spa->spa_evicting_os_list, os); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_deregister(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_remove(&spa->spa_evicting_os_list, os); cv_broadcast(&spa->spa_evicting_os_cv); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_wait(spa_t *spa) { mutex_enter(&spa->spa_evicting_os_lock); while (!list_is_empty(&spa->spa_evicting_os_list)) cv_wait(&spa->spa_evicting_os_cv, &spa->spa_evicting_os_lock); mutex_exit(&spa->spa_evicting_os_lock); dmu_buf_user_evict_wait(); } int spa_max_replication(spa_t *spa) { /* * As of SPA_VERSION == SPA_VERSION_DITTO_BLOCKS, we are able to * handle BPs with more than one DVA allocated. Set our max * replication level accordingly. */ if (spa_version(spa) < SPA_VERSION_DITTO_BLOCKS) return (1); return (MIN(SPA_DVAS_PER_BP, spa_max_replication_override)); } int spa_prev_software_version(spa_t *spa) { return (spa->spa_prev_software_version); } uint64_t spa_deadman_synctime(spa_t *spa) { return (spa->spa_deadman_synctime); } spa_autotrim_t spa_get_autotrim(spa_t *spa) { return (spa->spa_autotrim); } uint64_t spa_deadman_ziotime(spa_t *spa) { return (spa->spa_deadman_ziotime); } uint64_t spa_get_deadman_failmode(spa_t *spa) { return (spa->spa_deadman_failmode); } void spa_set_deadman_failmode(spa_t *spa, const char *failmode) { if (strcmp(failmode, "wait") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT; else if (strcmp(failmode, "continue") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_CONTINUE; else if (strcmp(failmode, "panic") == 0) spa->spa_deadman_failmode = ZIO_FAILURE_MODE_PANIC; else spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT; } void spa_set_deadman_ziotime(hrtime_t ns) { spa_t *spa = NULL; if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa->spa_deadman_ziotime = ns; mutex_exit(&spa_namespace_lock); } } void spa_set_deadman_synctime(hrtime_t ns) { spa_t *spa = NULL; if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa->spa_deadman_synctime = ns; mutex_exit(&spa_namespace_lock); } } uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva) { uint64_t asize = DVA_GET_ASIZE(dva); uint64_t dsize = asize; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (asize != 0 && spa->spa_deflate) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); if (vd != NULL) dsize = (asize >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio; } return (dsize); } uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); return (dsize); } uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); spa_config_exit(spa, SCL_VDEV, FTAG); return (dsize); } uint64_t spa_dirty_data(spa_t *spa) { return (spa->spa_dsl_pool->dp_dirty_total); } /* * ========================================================================== * SPA Import Progress Routines * ========================================================================== */ typedef struct spa_import_progress { uint64_t pool_guid; /* unique id for updates */ char *pool_name; spa_load_state_t spa_load_state; uint64_t mmp_sec_remaining; /* MMP activity check */ uint64_t spa_load_max_txg; /* rewind txg */ procfs_list_node_t smh_node; } spa_import_progress_t; spa_history_list_t *spa_import_progress_list = NULL; static int spa_import_progress_show_header(struct seq_file *f) { seq_printf(f, "%-20s %-14s %-14s %-12s %s\n", "pool_guid", "load_state", "multihost_secs", "max_txg", "pool_name"); return (0); } static int spa_import_progress_show(struct seq_file *f, void *data) { spa_import_progress_t *sip = (spa_import_progress_t *)data; seq_printf(f, "%-20llu %-14llu %-14llu %-12llu %s\n", (u_longlong_t)sip->pool_guid, (u_longlong_t)sip->spa_load_state, (u_longlong_t)sip->mmp_sec_remaining, (u_longlong_t)sip->spa_load_max_txg, (sip->pool_name ? sip->pool_name : "-")); return (0); } /* Remove oldest elements from list until there are no more than 'size' left */ static void spa_import_progress_truncate(spa_history_list_t *shl, unsigned int size) { spa_import_progress_t *sip; while (shl->size > size) { sip = list_remove_head(&shl->procfs_list.pl_list); if (sip->pool_name) spa_strfree(sip->pool_name); kmem_free(sip, sizeof (spa_import_progress_t)); shl->size--; } IMPLY(size == 0, list_is_empty(&shl->procfs_list.pl_list)); } static void spa_import_progress_init(void) { spa_import_progress_list = kmem_zalloc(sizeof (spa_history_list_t), KM_SLEEP); spa_import_progress_list->size = 0; spa_import_progress_list->procfs_list.pl_private = spa_import_progress_list; procfs_list_install("zfs", NULL, "import_progress", 0644, &spa_import_progress_list->procfs_list, spa_import_progress_show, spa_import_progress_show_header, NULL, offsetof(spa_import_progress_t, smh_node)); } static void spa_import_progress_destroy(void) { spa_history_list_t *shl = spa_import_progress_list; procfs_list_uninstall(&shl->procfs_list); spa_import_progress_truncate(shl, 0); procfs_list_destroy(&shl->procfs_list); kmem_free(shl, sizeof (spa_history_list_t)); } int spa_import_progress_set_state(uint64_t pool_guid, spa_load_state_t load_state) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->spa_load_state = load_state; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } int spa_import_progress_set_max_txg(uint64_t pool_guid, uint64_t load_max_txg) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->spa_load_max_txg = load_max_txg; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } int spa_import_progress_set_mmp_check(uint64_t pool_guid, uint64_t mmp_sec_remaining) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; int error = ENOENT; if (shl->size == 0) return (0); mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { sip->mmp_sec_remaining = mmp_sec_remaining; error = 0; break; } } mutex_exit(&shl->procfs_list.pl_lock); return (error); } /* * A new import is in progress, add an entry. */ void spa_import_progress_add(spa_t *spa) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; char *poolname = NULL; sip = kmem_zalloc(sizeof (spa_import_progress_t), KM_SLEEP); sip->pool_guid = spa_guid(spa); (void) nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &poolname); if (poolname == NULL) poolname = spa_name(spa); sip->pool_name = spa_strdup(poolname); sip->spa_load_state = spa_load_state(spa); mutex_enter(&shl->procfs_list.pl_lock); procfs_list_add(&shl->procfs_list, sip); shl->size++; mutex_exit(&shl->procfs_list.pl_lock); } void spa_import_progress_remove(uint64_t pool_guid) { spa_history_list_t *shl = spa_import_progress_list; spa_import_progress_t *sip; mutex_enter(&shl->procfs_list.pl_lock); for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL; sip = list_prev(&shl->procfs_list.pl_list, sip)) { if (sip->pool_guid == pool_guid) { if (sip->pool_name) spa_strfree(sip->pool_name); list_remove(&shl->procfs_list.pl_list, sip); shl->size--; kmem_free(sip, sizeof (spa_import_progress_t)); break; } } mutex_exit(&shl->procfs_list.pl_lock); } /* * ========================================================================== * Initialization and Termination * ========================================================================== */ static int spa_name_compare(const void *a1, const void *a2) { const spa_t *s1 = a1; const spa_t *s2 = a2; int s; s = strcmp(s1->spa_name, s2->spa_name); return (TREE_ISIGN(s)); } void spa_boot_init(void) { spa_config_load(); } void spa_init(spa_mode_t mode) { mutex_init(&spa_namespace_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_spare_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_l2cache_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa_namespace_cv, NULL, CV_DEFAULT, NULL); avl_create(&spa_namespace_avl, spa_name_compare, sizeof (spa_t), offsetof(spa_t, spa_avl)); avl_create(&spa_spare_avl, spa_spare_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); avl_create(&spa_l2cache_avl, spa_l2cache_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); spa_mode_global = mode; #ifndef _KERNEL if (spa_mode_global != SPA_MODE_READ && dprintf_find_string("watch")) { struct sigaction sa; sa.sa_flags = SA_SIGINFO; sigemptyset(&sa.sa_mask); sa.sa_sigaction = arc_buf_sigsegv; if (sigaction(SIGSEGV, &sa, NULL) == -1) { perror("could not enable watchpoints: " "sigaction(SIGSEGV, ...) = "); } else { arc_watch = B_TRUE; } } #endif fm_init(); zfs_refcount_init(); unique_init(); zfs_btree_init(); metaslab_stat_init(); ddt_init(); zio_init(); dmu_init(); zil_init(); vdev_cache_stat_init(); vdev_mirror_stat_init(); vdev_raidz_math_init(); vdev_file_init(); zfs_prop_init(); chksum_init(); zpool_prop_init(); zpool_feature_init(); spa_config_load(); vdev_prop_init(); l2arc_start(); scan_init(); qat_init(); spa_import_progress_init(); } void spa_fini(void) { l2arc_stop(); spa_evict_all(); vdev_file_fini(); vdev_cache_stat_fini(); vdev_mirror_stat_fini(); vdev_raidz_math_fini(); chksum_fini(); zil_fini(); dmu_fini(); zio_fini(); ddt_fini(); metaslab_stat_fini(); zfs_btree_fini(); unique_fini(); zfs_refcount_fini(); fm_fini(); scan_fini(); qat_fini(); spa_import_progress_destroy(); avl_destroy(&spa_namespace_avl); avl_destroy(&spa_spare_avl); avl_destroy(&spa_l2cache_avl); cv_destroy(&spa_namespace_cv); mutex_destroy(&spa_namespace_lock); mutex_destroy(&spa_spare_lock); mutex_destroy(&spa_l2cache_lock); } /* * Return whether this pool has a dedicated slog device. No locking needed. * It's not a problem if the wrong answer is returned as it's only for * performance and not correctness. */ boolean_t spa_has_slogs(spa_t *spa) { return (spa->spa_log_class->mc_groups != 0); } spa_log_state_t spa_get_log_state(spa_t *spa) { return (spa->spa_log_state); } void spa_set_log_state(spa_t *spa, spa_log_state_t state) { spa->spa_log_state = state; } boolean_t spa_is_root(spa_t *spa) { return (spa->spa_is_root); } boolean_t spa_writeable(spa_t *spa) { return (!!(spa->spa_mode & SPA_MODE_WRITE) && spa->spa_trust_config); } /* * Returns true if there is a pending sync task in any of the current * syncing txg, the current quiescing txg, or the current open txg. */ boolean_t spa_has_pending_synctask(spa_t *spa) { return (!txg_all_lists_empty(&spa->spa_dsl_pool->dp_sync_tasks) || !txg_all_lists_empty(&spa->spa_dsl_pool->dp_early_sync_tasks)); } spa_mode_t spa_mode(spa_t *spa) { return (spa->spa_mode); } uint64_t spa_bootfs(spa_t *spa) { return (spa->spa_bootfs); } uint64_t spa_delegation(spa_t *spa) { return (spa->spa_delegation); } objset_t * spa_meta_objset(spa_t *spa) { return (spa->spa_meta_objset); } enum zio_checksum spa_dedup_checksum(spa_t *spa) { return (spa->spa_dedup_checksum); } /* * Reset pool scan stat per scan pass (or reboot). */ void spa_scan_stat_init(spa_t *spa) { /* data not stored on disk */ spa->spa_scan_pass_start = gethrestime_sec(); if (dsl_scan_is_paused_scrub(spa->spa_dsl_pool->dp_scan)) spa->spa_scan_pass_scrub_pause = spa->spa_scan_pass_start; else spa->spa_scan_pass_scrub_pause = 0; spa->spa_scan_pass_scrub_spent_paused = 0; spa->spa_scan_pass_exam = 0; spa->spa_scan_pass_issued = 0; vdev_scan_stat_init(spa->spa_root_vdev); } /* * Get scan stats for zpool status reports */ int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps) { dsl_scan_t *scn = spa->spa_dsl_pool ? spa->spa_dsl_pool->dp_scan : NULL; if (scn == NULL || scn->scn_phys.scn_func == POOL_SCAN_NONE) return (SET_ERROR(ENOENT)); memset(ps, 0, sizeof (pool_scan_stat_t)); /* data stored on disk */ ps->pss_func = scn->scn_phys.scn_func; ps->pss_state = scn->scn_phys.scn_state; ps->pss_start_time = scn->scn_phys.scn_start_time; ps->pss_end_time = scn->scn_phys.scn_end_time; ps->pss_to_examine = scn->scn_phys.scn_to_examine; ps->pss_examined = scn->scn_phys.scn_examined; ps->pss_to_process = scn->scn_phys.scn_to_process; ps->pss_processed = scn->scn_phys.scn_processed; ps->pss_errors = scn->scn_phys.scn_errors; /* data not stored on disk */ ps->pss_pass_exam = spa->spa_scan_pass_exam; ps->pss_pass_start = spa->spa_scan_pass_start; ps->pss_pass_scrub_pause = spa->spa_scan_pass_scrub_pause; ps->pss_pass_scrub_spent_paused = spa->spa_scan_pass_scrub_spent_paused; ps->pss_pass_issued = spa->spa_scan_pass_issued; ps->pss_issued = scn->scn_issued_before_pass + spa->spa_scan_pass_issued; return (0); } int spa_maxblocksize(spa_t *spa) { if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) return (SPA_MAXBLOCKSIZE); else return (SPA_OLD_MAXBLOCKSIZE); } /* * Returns the txg that the last device removal completed. No indirect mappings * have been added since this txg. */ uint64_t spa_get_last_removal_txg(spa_t *spa) { uint64_t vdevid; uint64_t ret = -1ULL; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); /* * sr_prev_indirect_vdev is only modified while holding all the * config locks, so it is sufficient to hold SCL_VDEV as reader when * examining it. */ vdevid = spa->spa_removing_phys.sr_prev_indirect_vdev; while (vdevid != -1ULL) { vdev_t *vd = vdev_lookup_top(spa, vdevid); vdev_indirect_births_t *vib = vd->vdev_indirect_births; ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops); /* * If the removal did not remap any data, we don't care. */ if (vdev_indirect_births_count(vib) != 0) { ret = vdev_indirect_births_last_entry_txg(vib); break; } vdevid = vd->vdev_indirect_config.vic_prev_indirect_vdev; } spa_config_exit(spa, SCL_VDEV, FTAG); IMPLY(ret != -1ULL, spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)); return (ret); } int spa_maxdnodesize(spa_t *spa) { if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) return (DNODE_MAX_SIZE); else return (DNODE_MIN_SIZE); } boolean_t spa_multihost(spa_t *spa) { return (spa->spa_multihost ? B_TRUE : B_FALSE); } uint32_t spa_get_hostid(spa_t *spa) { return (spa->spa_hostid); } boolean_t spa_trust_config(spa_t *spa) { return (spa->spa_trust_config); } uint64_t spa_missing_tvds_allowed(spa_t *spa) { return (spa->spa_missing_tvds_allowed); } space_map_t * spa_syncing_log_sm(spa_t *spa) { return (spa->spa_syncing_log_sm); } void spa_set_missing_tvds(spa_t *spa, uint64_t missing) { spa->spa_missing_tvds = missing; } /* * Return the pool state string ("ONLINE", "DEGRADED", "SUSPENDED", etc). */ const char * spa_state_to_name(spa_t *spa) { ASSERT3P(spa, !=, NULL); /* * it is possible for the spa to exist, without root vdev * as the spa transitions during import/export */ vdev_t *rvd = spa->spa_root_vdev; if (rvd == NULL) { return ("TRANSITIONING"); } vdev_state_t state = rvd->vdev_state; vdev_aux_t aux = rvd->vdev_stat.vs_aux; if (spa_suspended(spa) && (spa_get_failmode(spa) != ZIO_FAILURE_MODE_CONTINUE)) return ("SUSPENDED"); switch (state) { case VDEV_STATE_CLOSED: case VDEV_STATE_OFFLINE: return ("OFFLINE"); case VDEV_STATE_REMOVED: return ("REMOVED"); case VDEV_STATE_CANT_OPEN: if (aux == VDEV_AUX_CORRUPT_DATA || aux == VDEV_AUX_BAD_LOG) return ("FAULTED"); else if (aux == VDEV_AUX_SPLIT_POOL) return ("SPLIT"); else return ("UNAVAIL"); case VDEV_STATE_FAULTED: return ("FAULTED"); case VDEV_STATE_DEGRADED: return ("DEGRADED"); case VDEV_STATE_HEALTHY: return ("ONLINE"); default: break; } return ("UNKNOWN"); } boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; for (uint64_t c = 0; c < rvd->vdev_children; c++) { if (!vdev_is_spacemap_addressable(rvd->vdev_child[c])) return (B_FALSE); } return (B_TRUE); } boolean_t spa_has_checkpoint(spa_t *spa) { return (spa->spa_checkpoint_txg != 0); } boolean_t spa_importing_readonly_checkpoint(spa_t *spa) { return ((spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT) && spa->spa_mode == SPA_MODE_READ); } uint64_t spa_min_claim_txg(spa_t *spa) { uint64_t checkpoint_txg = spa->spa_uberblock.ub_checkpoint_txg; if (checkpoint_txg != 0) return (checkpoint_txg + 1); return (spa->spa_first_txg); } /* * If there is a checkpoint, async destroys may consume more space from * the pool instead of freeing it. In an attempt to save the pool from * getting suspended when it is about to run out of space, we stop * processing async destroys. */ boolean_t spa_suspend_async_destroy(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t unreserved = dsl_pool_unreserved_space(dp, ZFS_SPACE_CHECK_EXTRA_RESERVED); uint64_t used = dsl_dir_phys(dp->dp_root_dir)->dd_used_bytes; uint64_t avail = (unreserved > used) ? (unreserved - used) : 0; if (spa_has_checkpoint(spa) && avail == 0) return (B_TRUE); return (B_FALSE); } #if defined(_KERNEL) int param_set_deadman_failmode_common(const char *val) { spa_t *spa = NULL; char *p; if (val == NULL) return (SET_ERROR(EINVAL)); if ((p = strchr(val, '\n')) != NULL) *p = '\0'; if (strcmp(val, "wait") != 0 && strcmp(val, "continue") != 0 && strcmp(val, "panic")) return (SET_ERROR(EINVAL)); if (spa_mode_global != SPA_MODE_UNINIT) { mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) spa_set_deadman_failmode(spa, val); mutex_exit(&spa_namespace_lock); } return (0); } #endif /* Namespace manipulation */ EXPORT_SYMBOL(spa_lookup); EXPORT_SYMBOL(spa_add); EXPORT_SYMBOL(spa_remove); EXPORT_SYMBOL(spa_next); /* Refcount functions */ EXPORT_SYMBOL(spa_open_ref); EXPORT_SYMBOL(spa_close); EXPORT_SYMBOL(spa_refcount_zero); /* Pool configuration lock */ EXPORT_SYMBOL(spa_config_tryenter); EXPORT_SYMBOL(spa_config_enter); EXPORT_SYMBOL(spa_config_exit); EXPORT_SYMBOL(spa_config_held); /* Pool vdev add/remove lock */ EXPORT_SYMBOL(spa_vdev_enter); EXPORT_SYMBOL(spa_vdev_exit); /* Pool vdev state change lock */ EXPORT_SYMBOL(spa_vdev_state_enter); EXPORT_SYMBOL(spa_vdev_state_exit); /* Accessor functions */ EXPORT_SYMBOL(spa_shutting_down); EXPORT_SYMBOL(spa_get_dsl); EXPORT_SYMBOL(spa_get_rootblkptr); EXPORT_SYMBOL(spa_set_rootblkptr); EXPORT_SYMBOL(spa_altroot); EXPORT_SYMBOL(spa_sync_pass); EXPORT_SYMBOL(spa_name); EXPORT_SYMBOL(spa_guid); EXPORT_SYMBOL(spa_last_synced_txg); EXPORT_SYMBOL(spa_first_txg); EXPORT_SYMBOL(spa_syncing_txg); EXPORT_SYMBOL(spa_version); EXPORT_SYMBOL(spa_state); EXPORT_SYMBOL(spa_load_state); EXPORT_SYMBOL(spa_freeze_txg); EXPORT_SYMBOL(spa_get_dspace); EXPORT_SYMBOL(spa_update_dspace); EXPORT_SYMBOL(spa_deflate); EXPORT_SYMBOL(spa_normal_class); EXPORT_SYMBOL(spa_log_class); EXPORT_SYMBOL(spa_special_class); EXPORT_SYMBOL(spa_preferred_class); EXPORT_SYMBOL(spa_max_replication); EXPORT_SYMBOL(spa_prev_software_version); EXPORT_SYMBOL(spa_get_failmode); EXPORT_SYMBOL(spa_suspended); EXPORT_SYMBOL(spa_bootfs); EXPORT_SYMBOL(spa_delegation); EXPORT_SYMBOL(spa_meta_objset); EXPORT_SYMBOL(spa_maxblocksize); EXPORT_SYMBOL(spa_maxdnodesize); /* Miscellaneous support routines */ EXPORT_SYMBOL(spa_guid_exists); EXPORT_SYMBOL(spa_strdup); EXPORT_SYMBOL(spa_strfree); EXPORT_SYMBOL(spa_generate_guid); EXPORT_SYMBOL(snprintf_blkptr); EXPORT_SYMBOL(spa_freeze); EXPORT_SYMBOL(spa_upgrade); EXPORT_SYMBOL(spa_evict_all); EXPORT_SYMBOL(spa_lookup_by_guid); EXPORT_SYMBOL(spa_has_spare); EXPORT_SYMBOL(dva_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize_sync); EXPORT_SYMBOL(bp_get_dsize); EXPORT_SYMBOL(spa_has_slogs); EXPORT_SYMBOL(spa_is_root); EXPORT_SYMBOL(spa_writeable); EXPORT_SYMBOL(spa_mode); EXPORT_SYMBOL(spa_namespace_lock); EXPORT_SYMBOL(spa_trust_config); EXPORT_SYMBOL(spa_missing_tvds_allowed); EXPORT_SYMBOL(spa_set_missing_tvds); EXPORT_SYMBOL(spa_state_to_name); EXPORT_SYMBOL(spa_importing_readonly_checkpoint); EXPORT_SYMBOL(spa_min_claim_txg); EXPORT_SYMBOL(spa_suspend_async_destroy); EXPORT_SYMBOL(spa_has_checkpoint); EXPORT_SYMBOL(spa_top_vdevs_spacemap_addressable); ZFS_MODULE_PARAM(zfs, zfs_, flags, UINT, ZMOD_RW, "Set additional debugging flags"); ZFS_MODULE_PARAM(zfs, zfs_, recover, INT, ZMOD_RW, "Set to attempt to recover from fatal errors"); ZFS_MODULE_PARAM(zfs, zfs_, free_leak_on_eio, INT, ZMOD_RW, "Set to ignore IO errors during free and permanently leak the space"); ZFS_MODULE_PARAM(zfs_deadman, zfs_deadman_, checktime_ms, ULONG, ZMOD_RW, "Dead I/O check interval in milliseconds"); ZFS_MODULE_PARAM(zfs_deadman, zfs_deadman_, enabled, INT, ZMOD_RW, "Enable deadman timer"); ZFS_MODULE_PARAM(zfs_spa, spa_, asize_inflation, UINT, ZMOD_RW, "SPA size estimate multiplication factor"); ZFS_MODULE_PARAM(zfs, zfs_, ddt_data_is_special, INT, ZMOD_RW, "Place DDT data into the special class"); ZFS_MODULE_PARAM(zfs, zfs_, user_indirect_is_special, INT, ZMOD_RW, "Place user data indirect blocks into the special class"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, failmode, param_set_deadman_failmode, param_get_charp, ZMOD_RW, "Failmode for deadman timer"); ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, synctime_ms, param_set_deadman_synctime, param_get_ulong, ZMOD_RW, "Pool sync expiration time in milliseconds"); ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, ziotime_ms, param_set_deadman_ziotime, param_get_ulong, ZMOD_RW, "IO expiration time in milliseconds"); ZFS_MODULE_PARAM(zfs, zfs_, special_class_metadata_reserve_pct, UINT, ZMOD_RW, "Small file blocks in special vdevs depends on this much " "free space available"); /* END CSTYLED */ ZFS_MODULE_PARAM_CALL(zfs_spa, spa_, slop_shift, param_set_slop_shift, param_get_uint, ZMOD_RW, "Reserved free space in pool"); diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c index 53c767e3bcb1..b097e09210af 100644 --- a/module/zfs/vdev.c +++ b/module/zfs/vdev.c @@ -1,6107 +1,6166 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2021 by Delphix. All rights reserved. * Copyright 2017 Nexenta Systems, Inc. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Toomas Soome * Copyright 2017 Joyent, Inc. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Datto Inc. All rights reserved. * Copyright (c) 2021, Klara Inc. * Copyright [2021] Hewlett Packard Enterprise Development LP */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_prop.h" /* * One metaslab from each (normal-class) vdev is used by the ZIL. These are * called "embedded slog metaslabs", are referenced by vdev_log_mg, and are * part of the spa_embedded_log_class. The metaslab with the most free space * in each vdev is selected for this purpose when the pool is opened (or a * vdev is added). See vdev_metaslab_init(). * * Log blocks can be allocated from the following locations. Each one is tried * in order until the allocation succeeds: * 1. dedicated log vdevs, aka "slog" (spa_log_class) * 2. embedded slog metaslabs (spa_embedded_log_class) * 3. other metaslabs in normal vdevs (spa_normal_class) * * zfs_embedded_slog_min_ms disables the embedded slog if there are fewer * than this number of metaslabs in the vdev. This ensures that we don't set * aside an unreasonable amount of space for the ZIL. If set to less than * 1 << (spa_slop_shift + 1), on small pools the usable space may be reduced * (by more than 1<vdev_path != NULL) { zfs_dbgmsg("%s vdev '%s': %s", vd->vdev_ops->vdev_op_type, vd->vdev_path, buf); } else { zfs_dbgmsg("%s-%llu vdev (guid %llu): %s", vd->vdev_ops->vdev_op_type, (u_longlong_t)vd->vdev_id, (u_longlong_t)vd->vdev_guid, buf); } } void vdev_dbgmsg_print_tree(vdev_t *vd, int indent) { char state[20]; if (vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops) { zfs_dbgmsg("%*svdev %llu: %s", indent, "", (u_longlong_t)vd->vdev_id, vd->vdev_ops->vdev_op_type); return; } switch (vd->vdev_state) { case VDEV_STATE_UNKNOWN: (void) snprintf(state, sizeof (state), "unknown"); break; case VDEV_STATE_CLOSED: (void) snprintf(state, sizeof (state), "closed"); break; case VDEV_STATE_OFFLINE: (void) snprintf(state, sizeof (state), "offline"); break; case VDEV_STATE_REMOVED: (void) snprintf(state, sizeof (state), "removed"); break; case VDEV_STATE_CANT_OPEN: (void) snprintf(state, sizeof (state), "can't open"); break; case VDEV_STATE_FAULTED: (void) snprintf(state, sizeof (state), "faulted"); break; case VDEV_STATE_DEGRADED: (void) snprintf(state, sizeof (state), "degraded"); break; case VDEV_STATE_HEALTHY: (void) snprintf(state, sizeof (state), "healthy"); break; default: (void) snprintf(state, sizeof (state), "", (uint_t)vd->vdev_state); } zfs_dbgmsg("%*svdev %u: %s%s, guid: %llu, path: %s, %s", indent, "", (int)vd->vdev_id, vd->vdev_ops->vdev_op_type, vd->vdev_islog ? " (log)" : "", (u_longlong_t)vd->vdev_guid, vd->vdev_path ? vd->vdev_path : "N/A", state); for (uint64_t i = 0; i < vd->vdev_children; i++) vdev_dbgmsg_print_tree(vd->vdev_child[i], indent + 2); } /* * Virtual device management. */ static const vdev_ops_t *const vdev_ops_table[] = { &vdev_root_ops, &vdev_raidz_ops, &vdev_draid_ops, &vdev_draid_spare_ops, &vdev_mirror_ops, &vdev_replacing_ops, &vdev_spare_ops, &vdev_disk_ops, &vdev_file_ops, &vdev_missing_ops, &vdev_hole_ops, &vdev_indirect_ops, NULL }; /* * Given a vdev type, return the appropriate ops vector. */ static vdev_ops_t * vdev_getops(const char *type) { const vdev_ops_t *ops, *const *opspp; for (opspp = vdev_ops_table; (ops = *opspp) != NULL; opspp++) if (strcmp(ops->vdev_op_type, type) == 0) break; return (ops); } /* * Given a vdev and a metaslab class, find which metaslab group we're * interested in. All vdevs may belong to two different metaslab classes. * Dedicated slog devices use only the primary metaslab group, rather than a * separate log group. For embedded slogs, the vdev_log_mg will be non-NULL. */ metaslab_group_t * vdev_get_mg(vdev_t *vd, metaslab_class_t *mc) { if (mc == spa_embedded_log_class(vd->vdev_spa) && vd->vdev_log_mg != NULL) return (vd->vdev_log_mg); else return (vd->vdev_mg); } void vdev_default_xlate(vdev_t *vd, const range_seg64_t *logical_rs, range_seg64_t *physical_rs, range_seg64_t *remain_rs) { (void) vd, (void) remain_rs; physical_rs->rs_start = logical_rs->rs_start; physical_rs->rs_end = logical_rs->rs_end; } /* * Derive the enumerated allocation bias from string input. * String origin is either the per-vdev zap or zpool(8). */ static vdev_alloc_bias_t vdev_derive_alloc_bias(const char *bias) { vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE; if (strcmp(bias, VDEV_ALLOC_BIAS_LOG) == 0) alloc_bias = VDEV_BIAS_LOG; else if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) alloc_bias = VDEV_BIAS_SPECIAL; else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0) alloc_bias = VDEV_BIAS_DEDUP; return (alloc_bias); } /* * Default asize function: return the MAX of psize with the asize of * all children. This is what's used by anything other than RAID-Z. */ uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize) { uint64_t asize = P2ROUNDUP(psize, 1ULL << vd->vdev_top->vdev_ashift); uint64_t csize; for (int c = 0; c < vd->vdev_children; c++) { csize = vdev_psize_to_asize(vd->vdev_child[c], psize); asize = MAX(asize, csize); } return (asize); } uint64_t vdev_default_min_asize(vdev_t *vd) { return (vd->vdev_min_asize); } /* * Get the minimum allocatable size. We define the allocatable size as * the vdev's asize rounded to the nearest metaslab. This allows us to * replace or attach devices which don't have the same physical size but * can still satisfy the same number of allocations. */ uint64_t vdev_get_min_asize(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; /* * If our parent is NULL (inactive spare or cache) or is the root, * just return our own asize. */ if (pvd == NULL) return (vd->vdev_asize); /* * The top-level vdev just returns the allocatable size rounded * to the nearest metaslab. */ if (vd == vd->vdev_top) return (P2ALIGN(vd->vdev_asize, 1ULL << vd->vdev_ms_shift)); return (pvd->vdev_ops->vdev_op_min_asize(pvd)); } void vdev_set_min_asize(vdev_t *vd) { vd->vdev_min_asize = vdev_get_min_asize(vd); for (int c = 0; c < vd->vdev_children; c++) vdev_set_min_asize(vd->vdev_child[c]); } /* * Get the minimal allocation size for the top-level vdev. */ uint64_t vdev_get_min_alloc(vdev_t *vd) { uint64_t min_alloc = 1ULL << vd->vdev_ashift; if (vd->vdev_ops->vdev_op_min_alloc != NULL) min_alloc = vd->vdev_ops->vdev_op_min_alloc(vd); return (min_alloc); } /* * Get the parity level for a top-level vdev. */ uint64_t vdev_get_nparity(vdev_t *vd) { uint64_t nparity = 0; if (vd->vdev_ops->vdev_op_nparity != NULL) nparity = vd->vdev_ops->vdev_op_nparity(vd); return (nparity); } /* * Get the number of data disks for a top-level vdev. */ uint64_t vdev_get_ndisks(vdev_t *vd) { uint64_t ndisks = 1; if (vd->vdev_ops->vdev_op_ndisks != NULL) ndisks = vd->vdev_ops->vdev_op_ndisks(vd); return (ndisks); } vdev_t * vdev_lookup_top(spa_t *spa, uint64_t vdev) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (vdev < rvd->vdev_children) { ASSERT(rvd->vdev_child[vdev] != NULL); return (rvd->vdev_child[vdev]); } return (NULL); } vdev_t * vdev_lookup_by_guid(vdev_t *vd, uint64_t guid) { vdev_t *mvd; if (vd->vdev_guid == guid) return (vd); for (int c = 0; c < vd->vdev_children; c++) if ((mvd = vdev_lookup_by_guid(vd->vdev_child[c], guid)) != NULL) return (mvd); return (NULL); } static int vdev_count_leaves_impl(vdev_t *vd) { int n = 0; if (vd->vdev_ops->vdev_op_leaf) return (1); for (int c = 0; c < vd->vdev_children; c++) n += vdev_count_leaves_impl(vd->vdev_child[c]); return (n); } int vdev_count_leaves(spa_t *spa) { int rc; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); rc = vdev_count_leaves_impl(spa->spa_root_vdev); spa_config_exit(spa, SCL_VDEV, FTAG); return (rc); } void vdev_add_child(vdev_t *pvd, vdev_t *cvd) { size_t oldsize, newsize; uint64_t id = cvd->vdev_id; vdev_t **newchild; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(cvd->vdev_parent == NULL); cvd->vdev_parent = pvd; if (pvd == NULL) return; ASSERT(id >= pvd->vdev_children || pvd->vdev_child[id] == NULL); oldsize = pvd->vdev_children * sizeof (vdev_t *); pvd->vdev_children = MAX(pvd->vdev_children, id + 1); newsize = pvd->vdev_children * sizeof (vdev_t *); newchild = kmem_alloc(newsize, KM_SLEEP); if (pvd->vdev_child != NULL) { memcpy(newchild, pvd->vdev_child, oldsize); kmem_free(pvd->vdev_child, oldsize); } pvd->vdev_child = newchild; pvd->vdev_child[id] = cvd; cvd->vdev_top = (pvd->vdev_top ? pvd->vdev_top: cvd); ASSERT(cvd->vdev_top->vdev_parent->vdev_parent == NULL); /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum += cvd->vdev_guid_sum; if (cvd->vdev_ops->vdev_op_leaf) { list_insert_head(&cvd->vdev_spa->spa_leaf_list, cvd); cvd->vdev_spa->spa_leaf_list_gen++; } } void vdev_remove_child(vdev_t *pvd, vdev_t *cvd) { int c; uint_t id = cvd->vdev_id; ASSERT(cvd->vdev_parent == pvd); if (pvd == NULL) return; ASSERT(id < pvd->vdev_children); ASSERT(pvd->vdev_child[id] == cvd); pvd->vdev_child[id] = NULL; cvd->vdev_parent = NULL; for (c = 0; c < pvd->vdev_children; c++) if (pvd->vdev_child[c]) break; if (c == pvd->vdev_children) { kmem_free(pvd->vdev_child, c * sizeof (vdev_t *)); pvd->vdev_child = NULL; pvd->vdev_children = 0; } if (cvd->vdev_ops->vdev_op_leaf) { spa_t *spa = cvd->vdev_spa; list_remove(&spa->spa_leaf_list, cvd); spa->spa_leaf_list_gen++; } /* * Walk up all ancestors to update guid sum. */ for (; pvd != NULL; pvd = pvd->vdev_parent) pvd->vdev_guid_sum -= cvd->vdev_guid_sum; } /* * Remove any holes in the child array. */ void vdev_compact_children(vdev_t *pvd) { vdev_t **newchild, *cvd; int oldc = pvd->vdev_children; int newc; ASSERT(spa_config_held(pvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (oldc == 0) return; for (int c = newc = 0; c < oldc; c++) if (pvd->vdev_child[c]) newc++; if (newc > 0) { newchild = kmem_zalloc(newc * sizeof (vdev_t *), KM_SLEEP); for (int c = newc = 0; c < oldc; c++) { if ((cvd = pvd->vdev_child[c]) != NULL) { newchild[newc] = cvd; cvd->vdev_id = newc++; } } } else { newchild = NULL; } kmem_free(pvd->vdev_child, oldc * sizeof (vdev_t *)); pvd->vdev_child = newchild; pvd->vdev_children = newc; } /* * Allocate and minimally initialize a vdev_t. */ vdev_t * vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops) { vdev_t *vd; vdev_indirect_config_t *vic; vd = kmem_zalloc(sizeof (vdev_t), KM_SLEEP); vic = &vd->vdev_indirect_config; if (spa->spa_root_vdev == NULL) { ASSERT(ops == &vdev_root_ops); spa->spa_root_vdev = vd; spa->spa_load_guid = spa_generate_guid(NULL); } if (guid == 0 && ops != &vdev_hole_ops) { if (spa->spa_root_vdev == vd) { /* * The root vdev's guid will also be the pool guid, * which must be unique among all pools. */ guid = spa_generate_guid(NULL); } else { /* * Any other vdev's guid must be unique within the pool. */ guid = spa_generate_guid(spa); } ASSERT(!spa_guid_exists(spa_guid(spa), guid)); } vd->vdev_spa = spa; vd->vdev_id = id; vd->vdev_guid = guid; vd->vdev_guid_sum = guid; vd->vdev_ops = ops; vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_ishole = (ops == &vdev_hole_ops); vic->vic_prev_indirect_vdev = UINT64_MAX; rw_init(&vd->vdev_indirect_rwlock, NULL, RW_DEFAULT, NULL); mutex_init(&vd->vdev_obsolete_lock, NULL, MUTEX_DEFAULT, NULL); vd->vdev_obsolete_segments = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); /* * Initialize rate limit structs for events. We rate limit ZIO delay * and checksum events so that we don't overwhelm ZED with thousands * of events when a disk is acting up. */ zfs_ratelimit_init(&vd->vdev_delay_rl, &zfs_slow_io_events_per_second, 1); zfs_ratelimit_init(&vd->vdev_deadman_rl, &zfs_slow_io_events_per_second, 1); zfs_ratelimit_init(&vd->vdev_checksum_rl, &zfs_checksum_events_per_second, 1); list_link_init(&vd->vdev_config_dirty_node); list_link_init(&vd->vdev_state_dirty_node); list_link_init(&vd->vdev_initialize_node); list_link_init(&vd->vdev_leaf_node); list_link_init(&vd->vdev_trim_node); mutex_init(&vd->vdev_dtl_lock, NULL, MUTEX_NOLOCKDEP, NULL); mutex_init(&vd->vdev_stat_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_scan_io_queue_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_initialize_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_initialize_io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_initialize_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_initialize_io_cv, NULL, CV_DEFAULT, NULL); mutex_init(&vd->vdev_trim_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_autotrim_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_trim_io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_trim_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_autotrim_cv, NULL, CV_DEFAULT, NULL); cv_init(&vd->vdev_trim_io_cv, NULL, CV_DEFAULT, NULL); mutex_init(&vd->vdev_rebuild_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&vd->vdev_rebuild_cv, NULL, CV_DEFAULT, NULL); for (int t = 0; t < DTL_TYPES; t++) { vd->vdev_dtl[t] = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); } txg_list_create(&vd->vdev_ms_list, spa, offsetof(struct metaslab, ms_txg_node)); txg_list_create(&vd->vdev_dtl_list, spa, offsetof(struct vdev, vdev_dtl_node)); vd->vdev_stat.vs_timestamp = gethrtime(); vdev_queue_init(vd); vdev_cache_init(vd); return (vd); } /* * Allocate a new vdev. The 'alloctype' is used to control whether we are * creating a new vdev or loading an existing one - the behavior is slightly * different for each case. */ int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int alloctype) { vdev_ops_t *ops; char *type; uint64_t guid = 0, islog; vdev_t *vd; vdev_indirect_config_t *vic; char *tmp = NULL; int rc; vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE; boolean_t top_level = (parent && !parent->vdev_parent); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0) return (SET_ERROR(EINVAL)); if ((ops = vdev_getops(type)) == NULL) return (SET_ERROR(EINVAL)); /* * If this is a load, get the vdev guid from the nvlist. * Otherwise, vdev_alloc_common() will generate one for us. */ if (alloctype == VDEV_ALLOC_LOAD) { uint64_t label_id; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &label_id) || label_id != id) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_SPARE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_L2CACHE) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } else if (alloctype == VDEV_ALLOC_ROOTPOOL) { if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0) return (SET_ERROR(EINVAL)); } /* * The first allocated vdev must be of type 'root'. */ if (ops != &vdev_root_ops && spa->spa_root_vdev == NULL) return (SET_ERROR(EINVAL)); /* * Determine whether we're a log vdev. */ islog = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &islog); if (islog && spa_version(spa) < SPA_VERSION_SLOGS) return (SET_ERROR(ENOTSUP)); if (ops == &vdev_hole_ops && spa_version(spa) < SPA_VERSION_HOLES) return (SET_ERROR(ENOTSUP)); if (top_level && alloctype == VDEV_ALLOC_ADD) { char *bias; /* * If creating a top-level vdev, check for allocation * classes input. */ if (nvlist_lookup_string(nv, ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0) { alloc_bias = vdev_derive_alloc_bias(bias); /* spa_vdev_add() expects feature to be enabled */ if (spa->spa_load_state != SPA_LOAD_CREATE && !spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)) { return (SET_ERROR(ENOTSUP)); } } /* spa_vdev_add() expects feature to be enabled */ if (ops == &vdev_draid_ops && spa->spa_load_state != SPA_LOAD_CREATE && !spa_feature_is_enabled(spa, SPA_FEATURE_DRAID)) { return (SET_ERROR(ENOTSUP)); } } /* * Initialize the vdev specific data. This is done before calling * vdev_alloc_common() since it may fail and this simplifies the * error reporting and cleanup code paths. */ void *tsd = NULL; if (ops->vdev_op_init != NULL) { rc = ops->vdev_op_init(spa, nv, &tsd); if (rc != 0) { return (rc); } } vd = vdev_alloc_common(spa, id, guid, ops); vd->vdev_tsd = tsd; vd->vdev_islog = islog; if (top_level && alloc_bias != VDEV_BIAS_NONE) vd->vdev_alloc_bias = alloc_bias; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &vd->vdev_path) == 0) vd->vdev_path = spa_strdup(vd->vdev_path); /* * ZPOOL_CONFIG_AUX_STATE = "external" means we previously forced a * fault on a vdev and want it to persist across imports (like with * zpool offline -f). */ rc = nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &tmp); if (rc == 0 && tmp != NULL && strcmp(tmp, "external") == 0) { vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL; vd->vdev_faulted = 1; vd->vdev_label_aux = VDEV_AUX_EXTERNAL; } if (nvlist_lookup_string(nv, ZPOOL_CONFIG_DEVID, &vd->vdev_devid) == 0) vd->vdev_devid = spa_strdup(vd->vdev_devid); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PHYS_PATH, &vd->vdev_physpath) == 0) vd->vdev_physpath = spa_strdup(vd->vdev_physpath); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH, &vd->vdev_enc_sysfs_path) == 0) vd->vdev_enc_sysfs_path = spa_strdup(vd->vdev_enc_sysfs_path); if (nvlist_lookup_string(nv, ZPOOL_CONFIG_FRU, &vd->vdev_fru) == 0) vd->vdev_fru = spa_strdup(vd->vdev_fru); /* * Set the whole_disk property. If it's not specified, leave the value * as -1. */ if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &vd->vdev_wholedisk) != 0) vd->vdev_wholedisk = -1ULL; vic = &vd->vdev_indirect_config; ASSERT0(vic->vic_mapping_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_OBJECT, &vic->vic_mapping_object); ASSERT0(vic->vic_births_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_BIRTHS, &vic->vic_births_object); ASSERT3U(vic->vic_prev_indirect_vdev, ==, UINT64_MAX); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_PREV_INDIRECT_VDEV, &vic->vic_prev_indirect_vdev); /* * Look for the 'not present' flag. This will only be set if the device * was not present at the time of import. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &vd->vdev_not_present); /* * Get the alignment requirement. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &vd->vdev_ashift); /* * Retrieve the vdev creation time. */ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, &vd->vdev_crtxg); /* * If we're a top-level vdev, try to load the allocation parameters. */ if (top_level && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_SPLIT)) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY, &vd->vdev_ms_array); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT, &vd->vdev_ms_shift); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASIZE, &vd->vdev_asize); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NONALLOCATING, &vd->vdev_noalloc); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVING, &vd->vdev_removing); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_TOP_ZAP, &vd->vdev_top_zap); } else { ASSERT0(vd->vdev_top_zap); } if (top_level && alloctype != VDEV_ALLOC_ATTACH) { ASSERT(alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_ADD || alloctype == VDEV_ALLOC_SPLIT || alloctype == VDEV_ALLOC_ROOTPOOL); /* Note: metaslab_group_create() is now deferred */ } if (vd->vdev_ops->vdev_op_leaf && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_SPLIT)) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_LEAF_ZAP, &vd->vdev_leaf_zap); } else { ASSERT0(vd->vdev_leaf_zap); } /* * If we're a leaf vdev, try to load the DTL object and other state. */ if (vd->vdev_ops->vdev_op_leaf && (alloctype == VDEV_ALLOC_LOAD || alloctype == VDEV_ALLOC_L2CACHE || alloctype == VDEV_ALLOC_ROOTPOOL)) { if (alloctype == VDEV_ALLOC_LOAD) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DTL, &vd->vdev_dtl_object); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_UNSPARE, &vd->vdev_unspare); } if (alloctype == VDEV_ALLOC_ROOTPOOL) { uint64_t spare = 0; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE, &spare) == 0 && spare) spa_spare_add(vd); } (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE, &vd->vdev_offline); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG, &vd->vdev_resilver_txg); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REBUILD_TXG, &vd->vdev_rebuild_txg); if (nvlist_exists(nv, ZPOOL_CONFIG_RESILVER_DEFER)) vdev_defer_resilver(vd); /* * In general, when importing a pool we want to ignore the * persistent fault state, as the diagnosis made on another * system may not be valid in the current context. The only * exception is if we forced a vdev to a persistently faulted * state with 'zpool offline -f'. The persistent fault will * remain across imports until cleared. * * Local vdevs will remain in the faulted state. */ if (spa_load_state(spa) == SPA_LOAD_OPEN || spa_load_state(spa) == SPA_LOAD_IMPORT) { (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED, &vd->vdev_faulted); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DEGRADED, &vd->vdev_degraded); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED, &vd->vdev_removed); if (vd->vdev_faulted || vd->vdev_degraded) { char *aux; vd->vdev_label_aux = VDEV_AUX_ERR_EXCEEDED; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &aux) == 0 && strcmp(aux, "external") == 0) vd->vdev_label_aux = VDEV_AUX_EXTERNAL; else vd->vdev_faulted = 0ULL; } } } /* * Add ourselves to the parent's list of children. */ vdev_add_child(parent, vd); *vdp = vd; return (0); } void vdev_free(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT3P(vd->vdev_initialize_thread, ==, NULL); ASSERT3P(vd->vdev_trim_thread, ==, NULL); ASSERT3P(vd->vdev_autotrim_thread, ==, NULL); ASSERT3P(vd->vdev_rebuild_thread, ==, NULL); /* * Scan queues are normally destroyed at the end of a scan. If the * queue exists here, that implies the vdev is being removed while * the scan is still running. */ if (vd->vdev_scan_io_queue != NULL) { mutex_enter(&vd->vdev_scan_io_queue_lock); dsl_scan_io_queue_destroy(vd->vdev_scan_io_queue); vd->vdev_scan_io_queue = NULL; mutex_exit(&vd->vdev_scan_io_queue_lock); } /* * vdev_free() implies closing the vdev first. This is simpler than * trying to ensure complicated semantics for all callers. */ vdev_close(vd); ASSERT(!list_link_active(&vd->vdev_config_dirty_node)); ASSERT(!list_link_active(&vd->vdev_state_dirty_node)); /* * Free all children. */ for (int c = 0; c < vd->vdev_children; c++) vdev_free(vd->vdev_child[c]); ASSERT(vd->vdev_child == NULL); ASSERT(vd->vdev_guid_sum == vd->vdev_guid); if (vd->vdev_ops->vdev_op_fini != NULL) vd->vdev_ops->vdev_op_fini(vd); /* * Discard allocation state. */ if (vd->vdev_mg != NULL) { vdev_metaslab_fini(vd); metaslab_group_destroy(vd->vdev_mg); vd->vdev_mg = NULL; } if (vd->vdev_log_mg != NULL) { ASSERT0(vd->vdev_ms_count); metaslab_group_destroy(vd->vdev_log_mg); vd->vdev_log_mg = NULL; } ASSERT0(vd->vdev_stat.vs_space); ASSERT0(vd->vdev_stat.vs_dspace); ASSERT0(vd->vdev_stat.vs_alloc); /* * Remove this vdev from its parent's child list. */ vdev_remove_child(vd->vdev_parent, vd); ASSERT(vd->vdev_parent == NULL); ASSERT(!list_link_active(&vd->vdev_leaf_node)); /* * Clean up vdev structure. */ vdev_queue_fini(vd); vdev_cache_fini(vd); if (vd->vdev_path) spa_strfree(vd->vdev_path); if (vd->vdev_devid) spa_strfree(vd->vdev_devid); if (vd->vdev_physpath) spa_strfree(vd->vdev_physpath); if (vd->vdev_enc_sysfs_path) spa_strfree(vd->vdev_enc_sysfs_path); if (vd->vdev_fru) spa_strfree(vd->vdev_fru); if (vd->vdev_isspare) spa_spare_remove(vd); if (vd->vdev_isl2cache) spa_l2cache_remove(vd); txg_list_destroy(&vd->vdev_ms_list); txg_list_destroy(&vd->vdev_dtl_list); mutex_enter(&vd->vdev_dtl_lock); space_map_close(vd->vdev_dtl_sm); for (int t = 0; t < DTL_TYPES; t++) { range_tree_vacate(vd->vdev_dtl[t], NULL, NULL); range_tree_destroy(vd->vdev_dtl[t]); } mutex_exit(&vd->vdev_dtl_lock); EQUIV(vd->vdev_indirect_births != NULL, vd->vdev_indirect_mapping != NULL); if (vd->vdev_indirect_births != NULL) { vdev_indirect_mapping_close(vd->vdev_indirect_mapping); vdev_indirect_births_close(vd->vdev_indirect_births); } if (vd->vdev_obsolete_sm != NULL) { ASSERT(vd->vdev_removing || vd->vdev_ops == &vdev_indirect_ops); space_map_close(vd->vdev_obsolete_sm); vd->vdev_obsolete_sm = NULL; } range_tree_destroy(vd->vdev_obsolete_segments); rw_destroy(&vd->vdev_indirect_rwlock); mutex_destroy(&vd->vdev_obsolete_lock); mutex_destroy(&vd->vdev_dtl_lock); mutex_destroy(&vd->vdev_stat_lock); mutex_destroy(&vd->vdev_probe_lock); mutex_destroy(&vd->vdev_scan_io_queue_lock); mutex_destroy(&vd->vdev_initialize_lock); mutex_destroy(&vd->vdev_initialize_io_lock); cv_destroy(&vd->vdev_initialize_io_cv); cv_destroy(&vd->vdev_initialize_cv); mutex_destroy(&vd->vdev_trim_lock); mutex_destroy(&vd->vdev_autotrim_lock); mutex_destroy(&vd->vdev_trim_io_lock); cv_destroy(&vd->vdev_trim_cv); cv_destroy(&vd->vdev_autotrim_cv); cv_destroy(&vd->vdev_trim_io_cv); mutex_destroy(&vd->vdev_rebuild_lock); cv_destroy(&vd->vdev_rebuild_cv); zfs_ratelimit_fini(&vd->vdev_delay_rl); zfs_ratelimit_fini(&vd->vdev_deadman_rl); zfs_ratelimit_fini(&vd->vdev_checksum_rl); if (vd == spa->spa_root_vdev) spa->spa_root_vdev = NULL; kmem_free(vd, sizeof (vdev_t)); } /* * Transfer top-level vdev state from svd to tvd. */ static void vdev_top_transfer(vdev_t *svd, vdev_t *tvd) { spa_t *spa = svd->vdev_spa; metaslab_t *msp; vdev_t *vd; int t; ASSERT(tvd == tvd->vdev_top); tvd->vdev_pending_fastwrite = svd->vdev_pending_fastwrite; tvd->vdev_ms_array = svd->vdev_ms_array; tvd->vdev_ms_shift = svd->vdev_ms_shift; tvd->vdev_ms_count = svd->vdev_ms_count; tvd->vdev_top_zap = svd->vdev_top_zap; svd->vdev_ms_array = 0; svd->vdev_ms_shift = 0; svd->vdev_ms_count = 0; svd->vdev_top_zap = 0; if (tvd->vdev_mg) ASSERT3P(tvd->vdev_mg, ==, svd->vdev_mg); if (tvd->vdev_log_mg) ASSERT3P(tvd->vdev_log_mg, ==, svd->vdev_log_mg); tvd->vdev_mg = svd->vdev_mg; tvd->vdev_log_mg = svd->vdev_log_mg; tvd->vdev_ms = svd->vdev_ms; svd->vdev_mg = NULL; svd->vdev_log_mg = NULL; svd->vdev_ms = NULL; if (tvd->vdev_mg != NULL) tvd->vdev_mg->mg_vd = tvd; if (tvd->vdev_log_mg != NULL) tvd->vdev_log_mg->mg_vd = tvd; tvd->vdev_checkpoint_sm = svd->vdev_checkpoint_sm; svd->vdev_checkpoint_sm = NULL; tvd->vdev_alloc_bias = svd->vdev_alloc_bias; svd->vdev_alloc_bias = VDEV_BIAS_NONE; tvd->vdev_stat.vs_alloc = svd->vdev_stat.vs_alloc; tvd->vdev_stat.vs_space = svd->vdev_stat.vs_space; tvd->vdev_stat.vs_dspace = svd->vdev_stat.vs_dspace; svd->vdev_stat.vs_alloc = 0; svd->vdev_stat.vs_space = 0; svd->vdev_stat.vs_dspace = 0; /* * State which may be set on a top-level vdev that's in the * process of being removed. */ ASSERT0(tvd->vdev_indirect_config.vic_births_object); ASSERT0(tvd->vdev_indirect_config.vic_mapping_object); ASSERT3U(tvd->vdev_indirect_config.vic_prev_indirect_vdev, ==, -1ULL); ASSERT3P(tvd->vdev_indirect_mapping, ==, NULL); ASSERT3P(tvd->vdev_indirect_births, ==, NULL); ASSERT3P(tvd->vdev_obsolete_sm, ==, NULL); ASSERT0(tvd->vdev_noalloc); ASSERT0(tvd->vdev_removing); ASSERT0(tvd->vdev_rebuilding); tvd->vdev_noalloc = svd->vdev_noalloc; tvd->vdev_removing = svd->vdev_removing; tvd->vdev_rebuilding = svd->vdev_rebuilding; tvd->vdev_rebuild_config = svd->vdev_rebuild_config; tvd->vdev_indirect_config = svd->vdev_indirect_config; tvd->vdev_indirect_mapping = svd->vdev_indirect_mapping; tvd->vdev_indirect_births = svd->vdev_indirect_births; range_tree_swap(&svd->vdev_obsolete_segments, &tvd->vdev_obsolete_segments); tvd->vdev_obsolete_sm = svd->vdev_obsolete_sm; svd->vdev_indirect_config.vic_mapping_object = 0; svd->vdev_indirect_config.vic_births_object = 0; svd->vdev_indirect_config.vic_prev_indirect_vdev = -1ULL; svd->vdev_indirect_mapping = NULL; svd->vdev_indirect_births = NULL; svd->vdev_obsolete_sm = NULL; svd->vdev_noalloc = 0; svd->vdev_removing = 0; svd->vdev_rebuilding = 0; for (t = 0; t < TXG_SIZE; t++) { while ((msp = txg_list_remove(&svd->vdev_ms_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_ms_list, msp, t); while ((vd = txg_list_remove(&svd->vdev_dtl_list, t)) != NULL) (void) txg_list_add(&tvd->vdev_dtl_list, vd, t); if (txg_list_remove_this(&spa->spa_vdev_txg_list, svd, t)) (void) txg_list_add(&spa->spa_vdev_txg_list, tvd, t); } if (list_link_active(&svd->vdev_config_dirty_node)) { vdev_config_clean(svd); vdev_config_dirty(tvd); } if (list_link_active(&svd->vdev_state_dirty_node)) { vdev_state_clean(svd); vdev_state_dirty(tvd); } tvd->vdev_deflate_ratio = svd->vdev_deflate_ratio; svd->vdev_deflate_ratio = 0; tvd->vdev_islog = svd->vdev_islog; svd->vdev_islog = 0; dsl_scan_io_queue_vdev_xfer(svd, tvd); } static void vdev_top_update(vdev_t *tvd, vdev_t *vd) { if (vd == NULL) return; vd->vdev_top = tvd; for (int c = 0; c < vd->vdev_children; c++) vdev_top_update(tvd, vd->vdev_child[c]); } /* * Add a mirror/replacing vdev above an existing vdev. There is no need to * call .vdev_op_init() since mirror/replacing vdevs do not have private state. */ vdev_t * vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops) { spa_t *spa = cvd->vdev_spa; vdev_t *pvd = cvd->vdev_parent; vdev_t *mvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); mvd = vdev_alloc_common(spa, cvd->vdev_id, 0, ops); mvd->vdev_asize = cvd->vdev_asize; mvd->vdev_min_asize = cvd->vdev_min_asize; mvd->vdev_max_asize = cvd->vdev_max_asize; mvd->vdev_psize = cvd->vdev_psize; mvd->vdev_ashift = cvd->vdev_ashift; mvd->vdev_logical_ashift = cvd->vdev_logical_ashift; mvd->vdev_physical_ashift = cvd->vdev_physical_ashift; mvd->vdev_state = cvd->vdev_state; mvd->vdev_crtxg = cvd->vdev_crtxg; vdev_remove_child(pvd, cvd); vdev_add_child(pvd, mvd); cvd->vdev_id = mvd->vdev_children; vdev_add_child(mvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (mvd == mvd->vdev_top) vdev_top_transfer(cvd, mvd); return (mvd); } /* * Remove a 1-way mirror/replacing vdev from the tree. */ void vdev_remove_parent(vdev_t *cvd) { vdev_t *mvd = cvd->vdev_parent; vdev_t *pvd = mvd->vdev_parent; ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(mvd->vdev_children == 1); ASSERT(mvd->vdev_ops == &vdev_mirror_ops || mvd->vdev_ops == &vdev_replacing_ops || mvd->vdev_ops == &vdev_spare_ops); cvd->vdev_ashift = mvd->vdev_ashift; cvd->vdev_logical_ashift = mvd->vdev_logical_ashift; cvd->vdev_physical_ashift = mvd->vdev_physical_ashift; vdev_remove_child(mvd, cvd); vdev_remove_child(pvd, mvd); /* * If cvd will replace mvd as a top-level vdev, preserve mvd's guid. * Otherwise, we could have detached an offline device, and when we * go to import the pool we'll think we have two top-level vdevs, * instead of a different version of the same top-level vdev. */ if (mvd->vdev_top == mvd) { uint64_t guid_delta = mvd->vdev_guid - cvd->vdev_guid; cvd->vdev_orig_guid = cvd->vdev_guid; cvd->vdev_guid += guid_delta; cvd->vdev_guid_sum += guid_delta; /* * If pool not set for autoexpand, we need to also preserve * mvd's asize to prevent automatic expansion of cvd. * Otherwise if we are adjusting the mirror by attaching and * detaching children of non-uniform sizes, the mirror could * autoexpand, unexpectedly requiring larger devices to * re-establish the mirror. */ if (!cvd->vdev_spa->spa_autoexpand) cvd->vdev_asize = mvd->vdev_asize; } cvd->vdev_id = mvd->vdev_id; vdev_add_child(pvd, cvd); vdev_top_update(cvd->vdev_top, cvd->vdev_top); if (cvd == cvd->vdev_top) vdev_top_transfer(mvd, cvd); ASSERT(mvd->vdev_children == 0); vdev_free(mvd); } void vdev_metaslab_group_create(vdev_t *vd) { spa_t *spa = vd->vdev_spa; /* * metaslab_group_create was delayed until allocation bias was available */ if (vd->vdev_mg == NULL) { metaslab_class_t *mc; if (vd->vdev_islog && vd->vdev_alloc_bias == VDEV_BIAS_NONE) vd->vdev_alloc_bias = VDEV_BIAS_LOG; ASSERT3U(vd->vdev_islog, ==, (vd->vdev_alloc_bias == VDEV_BIAS_LOG)); switch (vd->vdev_alloc_bias) { case VDEV_BIAS_LOG: mc = spa_log_class(spa); break; case VDEV_BIAS_SPECIAL: mc = spa_special_class(spa); break; case VDEV_BIAS_DEDUP: mc = spa_dedup_class(spa); break; default: mc = spa_normal_class(spa); } vd->vdev_mg = metaslab_group_create(mc, vd, spa->spa_alloc_count); if (!vd->vdev_islog) { vd->vdev_log_mg = metaslab_group_create( spa_embedded_log_class(spa), vd, 1); } /* * The spa ashift min/max only apply for the normal metaslab * class. Class destination is late binding so ashift boundary * setting had to wait until now. */ if (vd->vdev_top == vd && vd->vdev_ashift != 0 && mc == spa_normal_class(spa) && vd->vdev_aux == NULL) { if (vd->vdev_ashift > spa->spa_max_ashift) spa->spa_max_ashift = vd->vdev_ashift; if (vd->vdev_ashift < spa->spa_min_ashift) spa->spa_min_ashift = vd->vdev_ashift; uint64_t min_alloc = vdev_get_min_alloc(vd); if (min_alloc < spa->spa_min_alloc) spa->spa_min_alloc = min_alloc; } } } int vdev_metaslab_init(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; uint64_t oldc = vd->vdev_ms_count; uint64_t newc = vd->vdev_asize >> vd->vdev_ms_shift; metaslab_t **mspp; int error; boolean_t expanding = (oldc != 0); ASSERT(txg == 0 || spa_config_held(spa, SCL_ALLOC, RW_WRITER)); /* * This vdev is not being allocated from yet or is a hole. */ if (vd->vdev_ms_shift == 0) return (0); ASSERT(!vd->vdev_ishole); ASSERT(oldc <= newc); mspp = vmem_zalloc(newc * sizeof (*mspp), KM_SLEEP); if (expanding) { memcpy(mspp, vd->vdev_ms, oldc * sizeof (*mspp)); vmem_free(vd->vdev_ms, oldc * sizeof (*mspp)); } vd->vdev_ms = mspp; vd->vdev_ms_count = newc; for (uint64_t m = oldc; m < newc; m++) { uint64_t object = 0; /* * vdev_ms_array may be 0 if we are creating the "fake" * metaslabs for an indirect vdev for zdb's leak detection. * See zdb_leak_init(). */ if (txg == 0 && vd->vdev_ms_array != 0) { error = dmu_read(spa->spa_meta_objset, vd->vdev_ms_array, m * sizeof (uint64_t), sizeof (uint64_t), &object, DMU_READ_PREFETCH); if (error != 0) { vdev_dbgmsg(vd, "unable to read the metaslab " "array [error=%d]", error); return (error); } } error = metaslab_init(vd->vdev_mg, m, object, txg, &(vd->vdev_ms[m])); if (error != 0) { vdev_dbgmsg(vd, "metaslab_init failed [error=%d]", error); return (error); } } /* * Find the emptiest metaslab on the vdev and mark it for use for * embedded slog by moving it from the regular to the log metaslab * group. */ if (vd->vdev_mg->mg_class == spa_normal_class(spa) && vd->vdev_ms_count > zfs_embedded_slog_min_ms && avl_is_empty(&vd->vdev_log_mg->mg_metaslab_tree)) { uint64_t slog_msid = 0; uint64_t smallest = UINT64_MAX; /* * Note, we only search the new metaslabs, because the old * (pre-existing) ones may be active (e.g. have non-empty * range_tree's), and we don't move them to the new * metaslab_t. */ for (uint64_t m = oldc; m < newc; m++) { uint64_t alloc = space_map_allocated(vd->vdev_ms[m]->ms_sm); if (alloc < smallest) { slog_msid = m; smallest = alloc; } } metaslab_t *slog_ms = vd->vdev_ms[slog_msid]; /* * The metaslab was marked as dirty at the end of * metaslab_init(). Remove it from the dirty list so that we * can uninitialize and reinitialize it to the new class. */ if (txg != 0) { (void) txg_list_remove_this(&vd->vdev_ms_list, slog_ms, txg); } uint64_t sm_obj = space_map_object(slog_ms->ms_sm); metaslab_fini(slog_ms); VERIFY0(metaslab_init(vd->vdev_log_mg, slog_msid, sm_obj, txg, &vd->vdev_ms[slog_msid])); } if (txg == 0) spa_config_enter(spa, SCL_ALLOC, FTAG, RW_WRITER); /* * If the vdev is marked as non-allocating then don't * activate the metaslabs since we want to ensure that * no allocations are performed on this device. */ if (vd->vdev_noalloc) { /* track non-allocating vdev space */ spa->spa_nonallocating_dspace += spa_deflate(spa) ? vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space; } else if (!expanding) { metaslab_group_activate(vd->vdev_mg); if (vd->vdev_log_mg != NULL) metaslab_group_activate(vd->vdev_log_mg); } if (txg == 0) spa_config_exit(spa, SCL_ALLOC, FTAG); return (0); } void vdev_metaslab_fini(vdev_t *vd) { if (vd->vdev_checkpoint_sm != NULL) { ASSERT(spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_POOL_CHECKPOINT)); space_map_close(vd->vdev_checkpoint_sm); /* * Even though we close the space map, we need to set its * pointer to NULL. The reason is that vdev_metaslab_fini() * may be called multiple times for certain operations * (i.e. when destroying a pool) so we need to ensure that * this clause never executes twice. This logic is similar * to the one used for the vdev_ms clause below. */ vd->vdev_checkpoint_sm = NULL; } if (vd->vdev_ms != NULL) { metaslab_group_t *mg = vd->vdev_mg; metaslab_group_passivate(mg); if (vd->vdev_log_mg != NULL) { ASSERT(!vd->vdev_islog); metaslab_group_passivate(vd->vdev_log_mg); } uint64_t count = vd->vdev_ms_count; for (uint64_t m = 0; m < count; m++) { metaslab_t *msp = vd->vdev_ms[m]; if (msp != NULL) metaslab_fini(msp); } vmem_free(vd->vdev_ms, count * sizeof (metaslab_t *)); vd->vdev_ms = NULL; vd->vdev_ms_count = 0; for (int i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) { ASSERT0(mg->mg_histogram[i]); if (vd->vdev_log_mg != NULL) ASSERT0(vd->vdev_log_mg->mg_histogram[i]); } } ASSERT0(vd->vdev_ms_count); ASSERT3U(vd->vdev_pending_fastwrite, ==, 0); } typedef struct vdev_probe_stats { boolean_t vps_readable; boolean_t vps_writeable; int vps_flags; } vdev_probe_stats_t; static void vdev_probe_done(zio_t *zio) { spa_t *spa = zio->io_spa; vdev_t *vd = zio->io_vd; vdev_probe_stats_t *vps = zio->io_private; ASSERT(vd->vdev_probe_zio != NULL); if (zio->io_type == ZIO_TYPE_READ) { if (zio->io_error == 0) vps->vps_readable = 1; if (zio->io_error == 0 && spa_writeable(spa)) { zio_nowait(zio_write_phys(vd->vdev_probe_zio, vd, zio->io_offset, zio->io_size, zio->io_abd, ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_WRITE, vps->vps_flags, B_TRUE)); } else { abd_free(zio->io_abd); } } else if (zio->io_type == ZIO_TYPE_WRITE) { if (zio->io_error == 0) vps->vps_writeable = 1; abd_free(zio->io_abd); } else if (zio->io_type == ZIO_TYPE_NULL) { zio_t *pio; zio_link_t *zl; vd->vdev_cant_read |= !vps->vps_readable; vd->vdev_cant_write |= !vps->vps_writeable; if (vdev_readable(vd) && (vdev_writeable(vd) || !spa_writeable(spa))) { zio->io_error = 0; } else { ASSERT(zio->io_error != 0); vdev_dbgmsg(vd, "failed probe"); (void) zfs_ereport_post(FM_EREPORT_ZFS_PROBE_FAILURE, spa, vd, NULL, NULL, 0); zio->io_error = SET_ERROR(ENXIO); } mutex_enter(&vd->vdev_probe_lock); ASSERT(vd->vdev_probe_zio == zio); vd->vdev_probe_zio = NULL; mutex_exit(&vd->vdev_probe_lock); zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) if (!vdev_accessible(vd, pio)) pio->io_error = SET_ERROR(ENXIO); kmem_free(vps, sizeof (*vps)); } } /* * Determine whether this device is accessible. * * Read and write to several known locations: the pad regions of each * vdev label but the first, which we leave alone in case it contains * a VTOC. */ zio_t * vdev_probe(vdev_t *vd, zio_t *zio) { spa_t *spa = vd->vdev_spa; vdev_probe_stats_t *vps = NULL; zio_t *pio; ASSERT(vd->vdev_ops->vdev_op_leaf); /* * Don't probe the probe. */ if (zio && (zio->io_flags & ZIO_FLAG_PROBE)) return (NULL); /* * To prevent 'probe storms' when a device fails, we create * just one probe i/o at a time. All zios that want to probe * this vdev will become parents of the probe io. */ mutex_enter(&vd->vdev_probe_lock); if ((pio = vd->vdev_probe_zio) == NULL) { vps = kmem_zalloc(sizeof (*vps), KM_SLEEP); vps->vps_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_PROBE | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE | ZIO_FLAG_TRYHARD; if (spa_config_held(spa, SCL_ZIO, RW_WRITER)) { /* * vdev_cant_read and vdev_cant_write can only * transition from TRUE to FALSE when we have the * SCL_ZIO lock as writer; otherwise they can only * transition from FALSE to TRUE. This ensures that * any zio looking at these values can assume that * failures persist for the life of the I/O. That's * important because when a device has intermittent * connectivity problems, we want to ensure that * they're ascribed to the device (ENXIO) and not * the zio (EIO). * * Since we hold SCL_ZIO as writer here, clear both * values so the probe can reevaluate from first * principles. */ vps->vps_flags |= ZIO_FLAG_CONFIG_WRITER; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; } vd->vdev_probe_zio = pio = zio_null(NULL, spa, vd, vdev_probe_done, vps, vps->vps_flags | ZIO_FLAG_DONT_PROPAGATE); /* * We can't change the vdev state in this context, so we * kick off an async task to do it on our behalf. */ if (zio != NULL) { vd->vdev_probe_wanted = B_TRUE; spa_async_request(spa, SPA_ASYNC_PROBE); } } if (zio != NULL) zio_add_child(zio, pio); mutex_exit(&vd->vdev_probe_lock); if (vps == NULL) { ASSERT(zio != NULL); return (NULL); } for (int l = 1; l < VDEV_LABELS; l++) { zio_nowait(zio_read_phys(pio, vd, vdev_label_offset(vd->vdev_psize, l, offsetof(vdev_label_t, vl_be)), VDEV_PAD_SIZE, abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE), ZIO_CHECKSUM_OFF, vdev_probe_done, vps, ZIO_PRIORITY_SYNC_READ, vps->vps_flags, B_TRUE)); } if (zio == NULL) return (pio); zio_nowait(pio); return (NULL); } static void vdev_load_child(void *arg) { vdev_t *vd = arg; vd->vdev_load_error = vdev_load(vd); } static void vdev_open_child(void *arg) { vdev_t *vd = arg; vd->vdev_open_thread = curthread; vd->vdev_open_error = vdev_open(vd); vd->vdev_open_thread = NULL; } static boolean_t vdev_uses_zvols(vdev_t *vd) { #ifdef _KERNEL if (zvol_is_zvol(vd->vdev_path)) return (B_TRUE); #endif for (int c = 0; c < vd->vdev_children; c++) if (vdev_uses_zvols(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } /* * Returns B_TRUE if the passed child should be opened. */ static boolean_t vdev_default_open_children_func(vdev_t *vd) { (void) vd; return (B_TRUE); } /* * Open the requested child vdevs. If any of the leaf vdevs are using * a ZFS volume then do the opens in a single thread. This avoids a * deadlock when the current thread is holding the spa_namespace_lock. */ static void vdev_open_children_impl(vdev_t *vd, vdev_open_children_func_t *open_func) { int children = vd->vdev_children; taskq_t *tq = taskq_create("vdev_open", children, minclsyspri, children, children, TASKQ_PREPOPULATE); vd->vdev_nonrot = B_TRUE; for (int c = 0; c < children; c++) { vdev_t *cvd = vd->vdev_child[c]; if (open_func(cvd) == B_FALSE) continue; if (tq == NULL || vdev_uses_zvols(vd)) { cvd->vdev_open_error = vdev_open(cvd); } else { VERIFY(taskq_dispatch(tq, vdev_open_child, cvd, TQ_SLEEP) != TASKQID_INVALID); } vd->vdev_nonrot &= cvd->vdev_nonrot; } if (tq != NULL) { taskq_wait(tq); taskq_destroy(tq); } } /* * Open all child vdevs. */ void vdev_open_children(vdev_t *vd) { vdev_open_children_impl(vd, vdev_default_open_children_func); } /* * Conditionally open a subset of child vdevs. */ void vdev_open_children_subset(vdev_t *vd, vdev_open_children_func_t *open_func) { vdev_open_children_impl(vd, open_func); } /* * Compute the raidz-deflation ratio. Note, we hard-code * in 128k (1 << 17) because it is the "typical" blocksize. * Even though SPA_MAXBLOCKSIZE changed, this algorithm can not change, * otherwise it would inconsistently account for existing bp's. */ static void vdev_set_deflate_ratio(vdev_t *vd) { if (vd == vd->vdev_top && !vd->vdev_ishole && vd->vdev_ashift != 0) { vd->vdev_deflate_ratio = (1 << 17) / (vdev_psize_to_asize(vd, 1 << 17) >> SPA_MINBLOCKSHIFT); } } /* * Choose the best of two ashifts, preferring one between logical ashift * (absolute minimum) and administrator defined maximum, otherwise take * the biggest of the two. */ uint64_t vdev_best_ashift(uint64_t logical, uint64_t a, uint64_t b) { if (a > logical && a <= zfs_vdev_max_auto_ashift) { if (b <= logical || b > zfs_vdev_max_auto_ashift) return (a); else return (MAX(a, b)); } else if (b <= logical || b > zfs_vdev_max_auto_ashift) return (MAX(a, b)); return (b); } /* * Maximize performance by inflating the configured ashift for top level * vdevs to be as close to the physical ashift as possible while maintaining * administrator defined limits and ensuring it doesn't go below the * logical ashift. */ static void vdev_ashift_optimize(vdev_t *vd) { ASSERT(vd == vd->vdev_top); if (vd->vdev_ashift < vd->vdev_physical_ashift && vd->vdev_physical_ashift <= zfs_vdev_max_auto_ashift) { vd->vdev_ashift = MIN( MAX(zfs_vdev_max_auto_ashift, vd->vdev_ashift), MAX(zfs_vdev_min_auto_ashift, vd->vdev_physical_ashift)); } else { /* * If the logical and physical ashifts are the same, then * we ensure that the top-level vdev's ashift is not smaller * than our minimum ashift value. For the unusual case * where logical ashift > physical ashift, we can't cap * the calculated ashift based on max ashift as that * would cause failures. * We still check if we need to increase it to match * the min ashift. */ vd->vdev_ashift = MAX(zfs_vdev_min_auto_ashift, vd->vdev_ashift); } } /* * Prepare a virtual device for access. */ int vdev_open(vdev_t *vd) { spa_t *spa = vd->vdev_spa; int error; uint64_t osize = 0; uint64_t max_osize = 0; uint64_t asize, max_asize, psize; uint64_t logical_ashift = 0; uint64_t physical_ashift = 0; ASSERT(vd->vdev_open_thread == curthread || spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); ASSERT(vd->vdev_state == VDEV_STATE_CLOSED || vd->vdev_state == VDEV_STATE_CANT_OPEN || vd->vdev_state == VDEV_STATE_OFFLINE); vd->vdev_stat.vs_aux = VDEV_AUX_NONE; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vd->vdev_min_asize = vdev_get_min_asize(vd); /* * If this vdev is not removed, check its fault status. If it's * faulted, bail out of the open. */ if (!vd->vdev_removed && vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } else if (vd->vdev_offline) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, VDEV_AUX_NONE); return (SET_ERROR(ENXIO)); } error = vd->vdev_ops->vdev_op_open(vd, &osize, &max_osize, &logical_ashift, &physical_ashift); + + /* Keep the device in removed state if unplugged */ + if (error == ENOENT && vd->vdev_removed) { + vdev_set_state(vd, B_TRUE, VDEV_STATE_REMOVED, + VDEV_AUX_NONE); + return (error); + } + /* * Physical volume size should never be larger than its max size, unless * the disk has shrunk while we were reading it or the device is buggy * or damaged: either way it's not safe for use, bail out of the open. */ if (osize > max_osize) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_OPEN_FAILED); return (SET_ERROR(ENXIO)); } /* * Reset the vdev_reopening flag so that we actually close * the vdev on error. */ vd->vdev_reopening = B_FALSE; if (zio_injection_enabled && error == 0) error = zio_handle_device_injection(vd, NULL, SET_ERROR(ENXIO)); if (error) { if (vd->vdev_removed && vd->vdev_stat.vs_aux != VDEV_AUX_OPEN_FAILED) vd->vdev_removed = B_FALSE; if (vd->vdev_stat.vs_aux == VDEV_AUX_CHILDREN_OFFLINE) { vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, vd->vdev_stat.vs_aux); } else { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, vd->vdev_stat.vs_aux); } return (error); } vd->vdev_removed = B_FALSE; /* * Recheck the faulted flag now that we have confirmed that * the vdev is accessible. If we're faulted, bail. */ if (vd->vdev_faulted) { ASSERT(vd->vdev_children == 0); ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED || vd->vdev_label_aux == VDEV_AUX_EXTERNAL); vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, vd->vdev_label_aux); return (SET_ERROR(ENXIO)); } if (vd->vdev_degraded) { ASSERT(vd->vdev_children == 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_ERR_EXCEEDED); } else { vdev_set_state(vd, B_TRUE, VDEV_STATE_HEALTHY, 0); } /* * For hole or missing vdevs we just return success. */ if (vd->vdev_ishole || vd->vdev_ops == &vdev_missing_ops) return (0); for (int c = 0; c < vd->vdev_children; c++) { if (vd->vdev_child[c]->vdev_state != VDEV_STATE_HEALTHY) { vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE); break; } } osize = P2ALIGN(osize, (uint64_t)sizeof (vdev_label_t)); max_osize = P2ALIGN(max_osize, (uint64_t)sizeof (vdev_label_t)); if (vd->vdev_children == 0) { if (osize < SPA_MINDEVSIZE) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = osize; asize = osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); max_asize = max_osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE); } else { if (vd->vdev_parent != NULL && osize < SPA_MINDEVSIZE - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE)) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_TOO_SMALL); return (SET_ERROR(EOVERFLOW)); } psize = 0; asize = osize; max_asize = max_osize; } /* * If the vdev was expanded, record this so that we can re-create the * uberblock rings in labels {2,3}, during the next sync. */ if ((psize > vd->vdev_psize) && (vd->vdev_psize != 0)) vd->vdev_copy_uberblocks = B_TRUE; vd->vdev_psize = psize; /* * Make sure the allocatable size hasn't shrunk too much. */ if (asize < vd->vdev_min_asize) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (SET_ERROR(EINVAL)); } /* * We can always set the logical/physical ashift members since * their values are only used to calculate the vdev_ashift when * the device is first added to the config. These values should * not be used for anything else since they may change whenever * the device is reopened and we don't store them in the label. */ vd->vdev_physical_ashift = MAX(physical_ashift, vd->vdev_physical_ashift); vd->vdev_logical_ashift = MAX(logical_ashift, vd->vdev_logical_ashift); if (vd->vdev_asize == 0) { /* * This is the first-ever open, so use the computed values. * For compatibility, a different ashift can be requested. */ vd->vdev_asize = asize; vd->vdev_max_asize = max_asize; /* * If the vdev_ashift was not overridden at creation time, * then set it the logical ashift and optimize the ashift. */ if (vd->vdev_ashift == 0) { vd->vdev_ashift = vd->vdev_logical_ashift; if (vd->vdev_logical_ashift > ASHIFT_MAX) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_ASHIFT_TOO_BIG); return (SET_ERROR(EDOM)); } if (vd->vdev_top == vd) { vdev_ashift_optimize(vd); } } if (vd->vdev_ashift != 0 && (vd->vdev_ashift < ASHIFT_MIN || vd->vdev_ashift > ASHIFT_MAX)) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_ASHIFT); return (SET_ERROR(EDOM)); } } else { /* * Make sure the alignment required hasn't increased. */ if (vd->vdev_ashift > vd->vdev_top->vdev_ashift && vd->vdev_ops->vdev_op_leaf) { (void) zfs_ereport_post( FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT, spa, vd, NULL, NULL, 0); vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); return (SET_ERROR(EDOM)); } vd->vdev_max_asize = max_asize; } /* * If all children are healthy we update asize if either: * The asize has increased, due to a device expansion caused by dynamic * LUN growth or vdev replacement, and automatic expansion is enabled; * making the additional space available. * * The asize has decreased, due to a device shrink usually caused by a * vdev replace with a smaller device. This ensures that calculations * based of max_asize and asize e.g. esize are always valid. It's safe * to do this as we've already validated that asize is greater than * vdev_min_asize. */ if (vd->vdev_state == VDEV_STATE_HEALTHY && ((asize > vd->vdev_asize && (vd->vdev_expanding || spa->spa_autoexpand)) || (asize < vd->vdev_asize))) vd->vdev_asize = asize; vdev_set_min_asize(vd); /* * Ensure we can issue some IO before declaring the * vdev open for business. */ if (vd->vdev_ops->vdev_op_leaf && (error = zio_wait(vdev_probe(vd, NULL))) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED, VDEV_AUX_ERR_EXCEEDED); return (error); } /* * Track the minimum allocation size. */ if (vd->vdev_top == vd && vd->vdev_ashift != 0 && vd->vdev_islog == 0 && vd->vdev_aux == NULL) { uint64_t min_alloc = vdev_get_min_alloc(vd); if (min_alloc < spa->spa_min_alloc) spa->spa_min_alloc = min_alloc; } /* * If this is a leaf vdev, assess whether a resilver is needed. * But don't do this if we are doing a reopen for a scrub, since * this would just restart the scrub we are already doing. */ if (vd->vdev_ops->vdev_op_leaf && !spa->spa_scrub_reopen) dsl_scan_assess_vdev(spa->spa_dsl_pool, vd); return (0); } static void vdev_validate_child(void *arg) { vdev_t *vd = arg; vd->vdev_validate_thread = curthread; vd->vdev_validate_error = vdev_validate(vd); vd->vdev_validate_thread = NULL; } /* * Called once the vdevs are all opened, this routine validates the label * contents. This needs to be done before vdev_load() so that we don't * inadvertently do repair I/Os to the wrong device. * * This function will only return failure if one of the vdevs indicates that it * has since been destroyed or exported. This is only possible if * /etc/zfs/zpool.cache was readonly at the time. Otherwise, the vdev state * will be updated but the function will return 0. */ int vdev_validate(vdev_t *vd) { spa_t *spa = vd->vdev_spa; taskq_t *tq = NULL; nvlist_t *label; uint64_t guid = 0, aux_guid = 0, top_guid; uint64_t state; nvlist_t *nvl; uint64_t txg; int children = vd->vdev_children; if (vdev_validate_skip) return (0); if (children > 0) { tq = taskq_create("vdev_validate", children, minclsyspri, children, children, TASKQ_PREPOPULATE); } for (uint64_t c = 0; c < children; c++) { vdev_t *cvd = vd->vdev_child[c]; if (tq == NULL || vdev_uses_zvols(cvd)) { vdev_validate_child(cvd); } else { VERIFY(taskq_dispatch(tq, vdev_validate_child, cvd, TQ_SLEEP) != TASKQID_INVALID); } } if (tq != NULL) { taskq_wait(tq); taskq_destroy(tq); } for (int c = 0; c < children; c++) { int error = vd->vdev_child[c]->vdev_validate_error; if (error != 0) return (SET_ERROR(EBADF)); } /* * If the device has already failed, or was marked offline, don't do * any further validation. Otherwise, label I/O will fail and we will * overwrite the previous state. */ if (!vd->vdev_ops->vdev_op_leaf || !vdev_readable(vd)) return (0); /* * If we are performing an extreme rewind, we allow for a label that * was modified at a point after the current txg. * If config lock is not held do not check for the txg. spa_sync could * be updating the vdev's label before updating spa_last_synced_txg. */ if (spa->spa_extreme_rewind || spa_last_synced_txg(spa) == 0 || spa_config_held(spa, SCL_CONFIG, RW_WRITER) != SCL_CONFIG) txg = UINT64_MAX; else txg = spa_last_synced_txg(spa); if ((label = vdev_label_read_config(vd, txg)) == NULL) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_BAD_LABEL); vdev_dbgmsg(vd, "vdev_validate: failed reading config for " "txg %llu", (u_longlong_t)txg); return (0); } /* * Determine if this vdev has been split off into another * pool. If so, then refuse to open it. */ if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_SPLIT_GUID, &aux_guid) == 0 && aux_guid == spa_guid(spa)) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_SPLIT_POOL); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: vdev split into other pool"); return (0); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID, &guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_POOL_GUID); return (0); } /* * If config is not trusted then ignore the spa guid check. This is * necessary because if the machine crashed during a re-guid the new * guid might have been written to all of the vdev labels, but not the * cached config. The check will be performed again once we have the * trusted config from the MOS. */ if (spa->spa_trust_config && guid != spa_guid(spa)) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: vdev label pool_guid doesn't " "match config (%llu != %llu)", (u_longlong_t)guid, (u_longlong_t)spa_guid(spa)); return (0); } if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_ORIG_GUID, &aux_guid) != 0) aux_guid = 0; if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_GUID); return (0); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_TOP_GUID, &top_guid) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_TOP_GUID); return (0); } /* * If this vdev just became a top-level vdev because its sibling was * detached, it will have adopted the parent's vdev guid -- but the * label may or may not be on disk yet. Fortunately, either version * of the label will have the same top guid, so if we're a top-level * vdev, we can safely compare to that instead. * However, if the config comes from a cachefile that failed to update * after the detach, a top-level vdev will appear as a non top-level * vdev in the config. Also relax the constraints if we perform an * extreme rewind. * * If we split this vdev off instead, then we also check the * original pool's guid. We don't want to consider the vdev * corrupt if it is partway through a split operation. */ if (vd->vdev_guid != guid && vd->vdev_guid != aux_guid) { boolean_t mismatch = B_FALSE; if (spa->spa_trust_config && !spa->spa_extreme_rewind) { if (vd != vd->vdev_top || vd->vdev_guid != top_guid) mismatch = B_TRUE; } else { if (vd->vdev_guid != top_guid && vd->vdev_top->vdev_guid != guid) mismatch = B_TRUE; } if (mismatch) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: config guid " "doesn't match label guid"); vdev_dbgmsg(vd, "CONFIG: guid %llu, top_guid %llu", (u_longlong_t)vd->vdev_guid, (u_longlong_t)vd->vdev_top->vdev_guid); vdev_dbgmsg(vd, "LABEL: guid %llu, top_guid %llu, " "aux_guid %llu", (u_longlong_t)guid, (u_longlong_t)top_guid, (u_longlong_t)aux_guid); return (0); } } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label", ZPOOL_CONFIG_POOL_STATE); return (0); } nvlist_free(label); /* * If this is a verbatim import, no need to check the * state of the pool. */ if (!(spa->spa_import_flags & ZFS_IMPORT_VERBATIM) && spa_load_state(spa) == SPA_LOAD_OPEN && state != POOL_STATE_ACTIVE) { vdev_dbgmsg(vd, "vdev_validate: invalid pool state (%llu) " "for spa %s", (u_longlong_t)state, spa->spa_name); return (SET_ERROR(EBADF)); } /* * If we were able to open and validate a vdev that was * previously marked permanently unavailable, clear that state * now. */ if (vd->vdev_not_present) vd->vdev_not_present = 0; return (0); } static void vdev_copy_path_impl(vdev_t *svd, vdev_t *dvd) { char *old, *new; if (svd->vdev_path != NULL && dvd->vdev_path != NULL) { if (strcmp(svd->vdev_path, dvd->vdev_path) != 0) { zfs_dbgmsg("vdev_copy_path: vdev %llu: path changed " "from '%s' to '%s'", (u_longlong_t)dvd->vdev_guid, dvd->vdev_path, svd->vdev_path); spa_strfree(dvd->vdev_path); dvd->vdev_path = spa_strdup(svd->vdev_path); } } else if (svd->vdev_path != NULL) { dvd->vdev_path = spa_strdup(svd->vdev_path); zfs_dbgmsg("vdev_copy_path: vdev %llu: path set to '%s'", (u_longlong_t)dvd->vdev_guid, dvd->vdev_path); } /* * Our enclosure sysfs path may have changed between imports */ old = dvd->vdev_enc_sysfs_path; new = svd->vdev_enc_sysfs_path; if ((old != NULL && new == NULL) || (old == NULL && new != NULL) || ((old != NULL && new != NULL) && strcmp(new, old) != 0)) { zfs_dbgmsg("vdev_copy_path: vdev %llu: vdev_enc_sysfs_path " "changed from '%s' to '%s'", (u_longlong_t)dvd->vdev_guid, old, new); if (dvd->vdev_enc_sysfs_path) spa_strfree(dvd->vdev_enc_sysfs_path); if (svd->vdev_enc_sysfs_path) { dvd->vdev_enc_sysfs_path = spa_strdup( svd->vdev_enc_sysfs_path); } else { dvd->vdev_enc_sysfs_path = NULL; } } } /* * Recursively copy vdev paths from one vdev to another. Source and destination * vdev trees must have same geometry otherwise return error. Intended to copy * paths from userland config into MOS config. */ int vdev_copy_path_strict(vdev_t *svd, vdev_t *dvd) { if ((svd->vdev_ops == &vdev_missing_ops) || (svd->vdev_ishole && dvd->vdev_ishole) || (dvd->vdev_ops == &vdev_indirect_ops)) return (0); if (svd->vdev_ops != dvd->vdev_ops) { vdev_dbgmsg(svd, "vdev_copy_path: vdev type mismatch: %s != %s", svd->vdev_ops->vdev_op_type, dvd->vdev_ops->vdev_op_type); return (SET_ERROR(EINVAL)); } if (svd->vdev_guid != dvd->vdev_guid) { vdev_dbgmsg(svd, "vdev_copy_path: guids mismatch (%llu != " "%llu)", (u_longlong_t)svd->vdev_guid, (u_longlong_t)dvd->vdev_guid); return (SET_ERROR(EINVAL)); } if (svd->vdev_children != dvd->vdev_children) { vdev_dbgmsg(svd, "vdev_copy_path: children count mismatch: " "%llu != %llu", (u_longlong_t)svd->vdev_children, (u_longlong_t)dvd->vdev_children); return (SET_ERROR(EINVAL)); } for (uint64_t i = 0; i < svd->vdev_children; i++) { int error = vdev_copy_path_strict(svd->vdev_child[i], dvd->vdev_child[i]); if (error != 0) return (error); } if (svd->vdev_ops->vdev_op_leaf) vdev_copy_path_impl(svd, dvd); return (0); } static void vdev_copy_path_search(vdev_t *stvd, vdev_t *dvd) { ASSERT(stvd->vdev_top == stvd); ASSERT3U(stvd->vdev_id, ==, dvd->vdev_top->vdev_id); for (uint64_t i = 0; i < dvd->vdev_children; i++) { vdev_copy_path_search(stvd, dvd->vdev_child[i]); } if (!dvd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(dvd)) return; /* * The idea here is that while a vdev can shift positions within * a top vdev (when replacing, attaching mirror, etc.) it cannot * step outside of it. */ vdev_t *vd = vdev_lookup_by_guid(stvd, dvd->vdev_guid); if (vd == NULL || vd->vdev_ops != dvd->vdev_ops) return; ASSERT(vd->vdev_ops->vdev_op_leaf); vdev_copy_path_impl(vd, dvd); } /* * Recursively copy vdev paths from one root vdev to another. Source and * destination vdev trees may differ in geometry. For each destination leaf * vdev, search a vdev with the same guid and top vdev id in the source. * Intended to copy paths from userland config into MOS config. */ void vdev_copy_path_relaxed(vdev_t *srvd, vdev_t *drvd) { uint64_t children = MIN(srvd->vdev_children, drvd->vdev_children); ASSERT(srvd->vdev_ops == &vdev_root_ops); ASSERT(drvd->vdev_ops == &vdev_root_ops); for (uint64_t i = 0; i < children; i++) { vdev_copy_path_search(srvd->vdev_child[i], drvd->vdev_child[i]); } } /* * Close a virtual device. */ void vdev_close(vdev_t *vd) { vdev_t *pvd = vd->vdev_parent; spa_t *spa __maybe_unused = vd->vdev_spa; ASSERT(vd != NULL); ASSERT(vd->vdev_open_thread == curthread || spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* * If our parent is reopening, then we are as well, unless we are * going offline. */ if (pvd != NULL && pvd->vdev_reopening) vd->vdev_reopening = (pvd->vdev_reopening && !vd->vdev_offline); vd->vdev_ops->vdev_op_close(vd); vdev_cache_purge(vd); /* * We record the previous state before we close it, so that if we are * doing a reopen(), we don't generate FMA ereports if we notice that * it's still faulted. */ vd->vdev_prevstate = vd->vdev_state; if (vd->vdev_offline) vd->vdev_state = VDEV_STATE_OFFLINE; else vd->vdev_state = VDEV_STATE_CLOSED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } void vdev_hold(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_is_root(spa)); if (spa->spa_state == POOL_STATE_UNINITIALIZED) return; for (int c = 0; c < vd->vdev_children; c++) vdev_hold(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops->vdev_op_hold != NULL) vd->vdev_ops->vdev_op_hold(vd); } void vdev_rele(vdev_t *vd) { ASSERT(spa_is_root(vd->vdev_spa)); for (int c = 0; c < vd->vdev_children; c++) vdev_rele(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops->vdev_op_rele != NULL) vd->vdev_ops->vdev_op_rele(vd); } /* * Reopen all interior vdevs and any unopened leaves. We don't actually * reopen leaf vdevs which had previously been opened as they might deadlock * on the spa_config_lock. Instead we only obtain the leaf's physical size. * If the leaf has never been opened then open it, as usual. */ void vdev_reopen(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); /* set the reopening flag unless we're taking the vdev offline */ vd->vdev_reopening = !vd->vdev_offline; vdev_close(vd); (void) vdev_open(vd); /* * Call vdev_validate() here to make sure we have the same device. * Otherwise, a device with an invalid label could be successfully * opened in response to vdev_reopen(). */ if (vd->vdev_aux) { (void) vdev_validate_aux(vd); if (vdev_readable(vd) && vdev_writeable(vd) && vd->vdev_aux == &spa->spa_l2cache) { /* * In case the vdev is present we should evict all ARC * buffers and pointers to log blocks and reclaim their * space before restoring its contents to L2ARC. */ if (l2arc_vdev_present(vd)) { l2arc_rebuild_vdev(vd, B_TRUE); } else { l2arc_add_vdev(spa, vd); } spa_async_request(spa, SPA_ASYNC_L2CACHE_REBUILD); spa_async_request(spa, SPA_ASYNC_L2CACHE_TRIM); } } else { (void) vdev_validate(vd); } /* * Reassess parent vdev's health. */ vdev_propagate_state(vd); } int vdev_create(vdev_t *vd, uint64_t txg, boolean_t isreplacing) { int error; /* * Normally, partial opens (e.g. of a mirror) are allowed. * For a create, however, we want to fail the request if * there are any components we can't open. */ error = vdev_open(vd); if (error || vd->vdev_state != VDEV_STATE_HEALTHY) { vdev_close(vd); return (error ? error : SET_ERROR(ENXIO)); } /* * Recursively load DTLs and initialize all labels. */ if ((error = vdev_dtl_load(vd)) != 0 || (error = vdev_label_init(vd, txg, isreplacing ? VDEV_LABEL_REPLACE : VDEV_LABEL_CREATE)) != 0) { vdev_close(vd); return (error); } return (0); } void vdev_metaslab_set_size(vdev_t *vd) { uint64_t asize = vd->vdev_asize; uint64_t ms_count = asize >> zfs_vdev_default_ms_shift; uint64_t ms_shift; /* * There are two dimensions to the metaslab sizing calculation: * the size of the metaslab and the count of metaslabs per vdev. * * The default values used below are a good balance between memory * usage (larger metaslab size means more memory needed for loaded * metaslabs; more metaslabs means more memory needed for the * metaslab_t structs), metaslab load time (larger metaslabs take * longer to load), and metaslab sync time (more metaslabs means * more time spent syncing all of them). * * In general, we aim for zfs_vdev_default_ms_count (200) metaslabs. * The range of the dimensions are as follows: * * 2^29 <= ms_size <= 2^34 * 16 <= ms_count <= 131,072 * * On the lower end of vdev sizes, we aim for metaslabs sizes of * at least 512MB (2^29) to minimize fragmentation effects when * testing with smaller devices. However, the count constraint * of at least 16 metaslabs will override this minimum size goal. * * On the upper end of vdev sizes, we aim for a maximum metaslab * size of 16GB. However, we will cap the total count to 2^17 * metaslabs to keep our memory footprint in check and let the * metaslab size grow from there if that limit is hit. * * The net effect of applying above constrains is summarized below. * * vdev size metaslab count * --------------|----------------- * < 8GB ~16 * 8GB - 100GB one per 512MB * 100GB - 3TB ~200 * 3TB - 2PB one per 16GB * > 2PB ~131,072 * -------------------------------- * * Finally, note that all of the above calculate the initial * number of metaslabs. Expanding a top-level vdev will result * in additional metaslabs being allocated making it possible * to exceed the zfs_vdev_ms_count_limit. */ if (ms_count < zfs_vdev_min_ms_count) ms_shift = highbit64(asize / zfs_vdev_min_ms_count); else if (ms_count > zfs_vdev_default_ms_count) ms_shift = highbit64(asize / zfs_vdev_default_ms_count); else ms_shift = zfs_vdev_default_ms_shift; if (ms_shift < SPA_MAXBLOCKSHIFT) { ms_shift = SPA_MAXBLOCKSHIFT; } else if (ms_shift > zfs_vdev_max_ms_shift) { ms_shift = zfs_vdev_max_ms_shift; /* cap the total count to constrain memory footprint */ if ((asize >> ms_shift) > zfs_vdev_ms_count_limit) ms_shift = highbit64(asize / zfs_vdev_ms_count_limit); } vd->vdev_ms_shift = ms_shift; ASSERT3U(vd->vdev_ms_shift, >=, SPA_MAXBLOCKSHIFT); } void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg) { ASSERT(vd == vd->vdev_top); /* indirect vdevs don't have metaslabs or dtls */ ASSERT(vdev_is_concrete(vd) || flags == 0); ASSERT(ISP2(flags)); ASSERT(spa_writeable(vd->vdev_spa)); if (flags & VDD_METASLAB) (void) txg_list_add(&vd->vdev_ms_list, arg, txg); if (flags & VDD_DTL) (void) txg_list_add(&vd->vdev_dtl_list, arg, txg); (void) txg_list_add(&vd->vdev_spa->spa_vdev_txg_list, vd, txg); } void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg) { for (int c = 0; c < vd->vdev_children; c++) vdev_dirty_leaves(vd->vdev_child[c], flags, txg); if (vd->vdev_ops->vdev_op_leaf) vdev_dirty(vd->vdev_top, flags, vd, txg); } /* * DTLs. * * A vdev's DTL (dirty time log) is the set of transaction groups for which * the vdev has less than perfect replication. There are four kinds of DTL: * * DTL_MISSING: txgs for which the vdev has no valid copies of the data * * DTL_PARTIAL: txgs for which data is available, but not fully replicated * * DTL_SCRUB: the txgs that could not be repaired by the last scrub; upon * scrub completion, DTL_SCRUB replaces DTL_MISSING in the range of * txgs that was scrubbed. * * DTL_OUTAGE: txgs which cannot currently be read, whether due to * persistent errors or just some device being offline. * Unlike the other three, the DTL_OUTAGE map is not generally * maintained; it's only computed when needed, typically to * determine whether a device can be detached. * * For leaf vdevs, DTL_MISSING and DTL_PARTIAL are identical: the device * either has the data or it doesn't. * * For interior vdevs such as mirror and RAID-Z the picture is more complex. * A vdev's DTL_PARTIAL is the union of its children's DTL_PARTIALs, because * if any child is less than fully replicated, then so is its parent. * A vdev's DTL_MISSING is a modified union of its children's DTL_MISSINGs, * comprising only those txgs which appear in 'maxfaults' or more children; * those are the txgs we don't have enough replication to read. For example, * double-parity RAID-Z can tolerate up to two missing devices (maxfaults == 2); * thus, its DTL_MISSING consists of the set of txgs that appear in more than * two child DTL_MISSING maps. * * It should be clear from the above that to compute the DTLs and outage maps * for all vdevs, it suffices to know just the leaf vdevs' DTL_MISSING maps. * Therefore, that is all we keep on disk. When loading the pool, or after * a configuration change, we generate all other DTLs from first principles. */ void vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); ASSERT(spa_writeable(vd->vdev_spa)); mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_contains(rt, txg, size)) range_tree_add(rt, txg, size); mutex_exit(&vd->vdev_dtl_lock); } boolean_t vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t dirty = B_FALSE; ASSERT(t < DTL_TYPES); ASSERT(vd != vd->vdev_spa->spa_root_vdev); /* * While we are loading the pool, the DTLs have not been loaded yet. * This isn't a problem but it can result in devices being tried * which are known to not have the data. In which case, the import * is relying on the checksum to ensure that we get the right data. * Note that while importing we are only reading the MOS, which is * always checksummed. */ mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_is_empty(rt)) dirty = range_tree_contains(rt, txg, size); mutex_exit(&vd->vdev_dtl_lock); return (dirty); } boolean_t vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t t) { range_tree_t *rt = vd->vdev_dtl[t]; boolean_t empty; mutex_enter(&vd->vdev_dtl_lock); empty = range_tree_is_empty(rt); mutex_exit(&vd->vdev_dtl_lock); return (empty); } /* * Check if the txg falls within the range which must be * resilvered. DVAs outside this range can always be skipped. */ boolean_t vdev_default_need_resilver(vdev_t *vd, const dva_t *dva, size_t psize, uint64_t phys_birth) { (void) dva, (void) psize; /* Set by sequential resilver. */ if (phys_birth == TXG_UNKNOWN) return (B_TRUE); return (vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1)); } /* * Returns B_TRUE if the vdev determines the DVA needs to be resilvered. */ boolean_t vdev_dtl_need_resilver(vdev_t *vd, const dva_t *dva, size_t psize, uint64_t phys_birth) { ASSERT(vd != vd->vdev_spa->spa_root_vdev); if (vd->vdev_ops->vdev_op_need_resilver == NULL || vd->vdev_ops->vdev_op_leaf) return (B_TRUE); return (vd->vdev_ops->vdev_op_need_resilver(vd, dva, psize, phys_birth)); } /* * Returns the lowest txg in the DTL range. */ static uint64_t vdev_dtl_min(vdev_t *vd) { ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); return (range_tree_min(vd->vdev_dtl[DTL_MISSING]) - 1); } /* * Returns the highest txg in the DTL. */ static uint64_t vdev_dtl_max(vdev_t *vd) { ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock)); ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0); ASSERT0(vd->vdev_children); return (range_tree_max(vd->vdev_dtl[DTL_MISSING])); } /* * Determine if a resilvering vdev should remove any DTL entries from * its range. If the vdev was resilvering for the entire duration of the * scan then it should excise that range from its DTLs. Otherwise, this * vdev is considered partially resilvered and should leave its DTL * entries intact. The comment in vdev_dtl_reassess() describes how we * excise the DTLs. */ static boolean_t vdev_dtl_should_excise(vdev_t *vd, boolean_t rebuild_done) { ASSERT0(vd->vdev_children); if (vd->vdev_state < VDEV_STATE_DEGRADED) return (B_FALSE); if (vd->vdev_resilver_deferred) return (B_FALSE); if (range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) return (B_TRUE); if (rebuild_done) { vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config; vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys; /* Rebuild not initiated by attach */ if (vd->vdev_rebuild_txg == 0) return (B_TRUE); /* * When a rebuild completes without error then all missing data * up to the rebuild max txg has been reconstructed and the DTL * is eligible for excision. */ if (vrp->vrp_rebuild_state == VDEV_REBUILD_COMPLETE && vdev_dtl_max(vd) <= vrp->vrp_max_txg) { ASSERT3U(vrp->vrp_min_txg, <=, vdev_dtl_min(vd)); ASSERT3U(vrp->vrp_min_txg, <, vd->vdev_rebuild_txg); ASSERT3U(vd->vdev_rebuild_txg, <=, vrp->vrp_max_txg); return (B_TRUE); } } else { dsl_scan_t *scn = vd->vdev_spa->spa_dsl_pool->dp_scan; dsl_scan_phys_t *scnp __maybe_unused = &scn->scn_phys; /* Resilver not initiated by attach */ if (vd->vdev_resilver_txg == 0) return (B_TRUE); /* * When a resilver is initiated the scan will assign the * scn_max_txg value to the highest txg value that exists * in all DTLs. If this device's max DTL is not part of this * scan (i.e. it is not in the range (scn_min_txg, scn_max_txg] * then it is not eligible for excision. */ if (vdev_dtl_max(vd) <= scn->scn_phys.scn_max_txg) { ASSERT3U(scnp->scn_min_txg, <=, vdev_dtl_min(vd)); ASSERT3U(scnp->scn_min_txg, <, vd->vdev_resilver_txg); ASSERT3U(vd->vdev_resilver_txg, <=, scnp->scn_max_txg); return (B_TRUE); } } return (B_FALSE); } /* * Reassess DTLs after a config change or scrub completion. If txg == 0 no * write operations will be issued to the pool. */ void vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg, boolean_t scrub_done, boolean_t rebuild_done) { spa_t *spa = vd->vdev_spa; avl_tree_t reftree; int minref; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); for (int c = 0; c < vd->vdev_children; c++) vdev_dtl_reassess(vd->vdev_child[c], txg, scrub_txg, scrub_done, rebuild_done); if (vd == spa->spa_root_vdev || !vdev_is_concrete(vd) || vd->vdev_aux) return; if (vd->vdev_ops->vdev_op_leaf) { dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config; boolean_t check_excise = B_FALSE; boolean_t wasempty = B_TRUE; mutex_enter(&vd->vdev_dtl_lock); /* * If requested, pretend the scan or rebuild completed cleanly. */ if (zfs_scan_ignore_errors) { if (scn != NULL) scn->scn_phys.scn_errors = 0; if (vr != NULL) vr->vr_rebuild_phys.vrp_errors = 0; } if (scrub_txg != 0 && !range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) { wasempty = B_FALSE; zfs_dbgmsg("guid:%llu txg:%llu scrub:%llu started:%d " "dtl:%llu/%llu errors:%llu", (u_longlong_t)vd->vdev_guid, (u_longlong_t)txg, (u_longlong_t)scrub_txg, spa->spa_scrub_started, (u_longlong_t)vdev_dtl_min(vd), (u_longlong_t)vdev_dtl_max(vd), (u_longlong_t)(scn ? scn->scn_phys.scn_errors : 0)); } /* * If we've completed a scrub/resilver or a rebuild cleanly * then determine if this vdev should remove any DTLs. We * only want to excise regions on vdevs that were available * during the entire duration of this scan. */ if (rebuild_done && vr != NULL && vr->vr_rebuild_phys.vrp_errors == 0) { check_excise = B_TRUE; } else { if (spa->spa_scrub_started || (scn != NULL && scn->scn_phys.scn_errors == 0)) { check_excise = B_TRUE; } } if (scrub_txg && check_excise && vdev_dtl_should_excise(vd, rebuild_done)) { /* * We completed a scrub, resilver or rebuild up to * scrub_txg. If we did it without rebooting, then * the scrub dtl will be valid, so excise the old * region and fold in the scrub dtl. Otherwise, * leave the dtl as-is if there was an error. * * There's little trick here: to excise the beginning * of the DTL_MISSING map, we put it into a reference * tree and then add a segment with refcnt -1 that * covers the range [0, scrub_txg). This means * that each txg in that range has refcnt -1 or 0. * We then add DTL_SCRUB with a refcnt of 2, so that * entries in the range [0, scrub_txg) will have a * positive refcnt -- either 1 or 2. We then convert * the reference tree into the new DTL_MISSING map. */ space_reftree_create(&reftree); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_add_seg(&reftree, 0, scrub_txg, -1); space_reftree_add_map(&reftree, vd->vdev_dtl[DTL_SCRUB], 2); space_reftree_generate_map(&reftree, vd->vdev_dtl[DTL_MISSING], 1); space_reftree_destroy(&reftree); if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) { zfs_dbgmsg("update DTL_MISSING:%llu/%llu", (u_longlong_t)vdev_dtl_min(vd), (u_longlong_t)vdev_dtl_max(vd)); } else if (!wasempty) { zfs_dbgmsg("DTL_MISSING is now empty"); } } range_tree_vacate(vd->vdev_dtl[DTL_PARTIAL], NULL, NULL); range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_PARTIAL]); if (scrub_done) range_tree_vacate(vd->vdev_dtl[DTL_SCRUB], NULL, NULL); range_tree_vacate(vd->vdev_dtl[DTL_OUTAGE], NULL, NULL); if (!vdev_readable(vd)) range_tree_add(vd->vdev_dtl[DTL_OUTAGE], 0, -1ULL); else range_tree_walk(vd->vdev_dtl[DTL_MISSING], range_tree_add, vd->vdev_dtl[DTL_OUTAGE]); /* * If the vdev was resilvering or rebuilding and no longer * has any DTLs then reset the appropriate flag and dirty * the top level so that we persist the change. */ if (txg != 0 && range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) && range_tree_is_empty(vd->vdev_dtl[DTL_OUTAGE])) { if (vd->vdev_rebuild_txg != 0) { vd->vdev_rebuild_txg = 0; vdev_config_dirty(vd->vdev_top); } else if (vd->vdev_resilver_txg != 0) { vd->vdev_resilver_txg = 0; vdev_config_dirty(vd->vdev_top); } } mutex_exit(&vd->vdev_dtl_lock); if (txg != 0) vdev_dirty(vd->vdev_top, VDD_DTL, vd, txg); return; } mutex_enter(&vd->vdev_dtl_lock); for (int t = 0; t < DTL_TYPES; t++) { /* account for child's outage in parent's missing map */ int s = (t == DTL_MISSING) ? DTL_OUTAGE: t; if (t == DTL_SCRUB) continue; /* leaf vdevs only */ if (t == DTL_PARTIAL) minref = 1; /* i.e. non-zero */ else if (vdev_get_nparity(vd) != 0) minref = vdev_get_nparity(vd) + 1; /* RAID-Z, dRAID */ else minref = vd->vdev_children; /* any kind of mirror */ space_reftree_create(&reftree); for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; mutex_enter(&cvd->vdev_dtl_lock); space_reftree_add_map(&reftree, cvd->vdev_dtl[s], 1); mutex_exit(&cvd->vdev_dtl_lock); } space_reftree_generate_map(&reftree, vd->vdev_dtl[t], minref); space_reftree_destroy(&reftree); } mutex_exit(&vd->vdev_dtl_lock); } +/* + * Iterate over all the vdevs except spare, and post kobj events + */ +void +vdev_post_kobj_evt(vdev_t *vd) +{ + if (vd->vdev_ops->vdev_op_kobj_evt_post && + vd->vdev_kobj_flag == B_FALSE) { + vd->vdev_kobj_flag = B_TRUE; + vd->vdev_ops->vdev_op_kobj_evt_post(vd); + } + + for (int c = 0; c < vd->vdev_children; c++) + vdev_post_kobj_evt(vd->vdev_child[c]); +} + +/* + * Iterate over all the vdevs except spare, and clear kobj events + */ +void +vdev_clear_kobj_evt(vdev_t *vd) +{ + vd->vdev_kobj_flag = B_FALSE; + + for (int c = 0; c < vd->vdev_children; c++) + vdev_clear_kobj_evt(vd->vdev_child[c]); +} + int vdev_dtl_load(vdev_t *vd) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; range_tree_t *rt; int error = 0; if (vd->vdev_ops->vdev_op_leaf && vd->vdev_dtl_object != 0) { ASSERT(vdev_is_concrete(vd)); /* * If the dtl cannot be sync'd there is no need to open it. */ if (spa->spa_mode == SPA_MODE_READ && !spa->spa_read_spacemaps) return (0); error = space_map_open(&vd->vdev_dtl_sm, mos, vd->vdev_dtl_object, 0, -1ULL, 0); if (error) return (error); ASSERT(vd->vdev_dtl_sm != NULL); rt = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); error = space_map_load(vd->vdev_dtl_sm, rt, SM_ALLOC); if (error == 0) { mutex_enter(&vd->vdev_dtl_lock); range_tree_walk(rt, range_tree_add, vd->vdev_dtl[DTL_MISSING]); mutex_exit(&vd->vdev_dtl_lock); } range_tree_vacate(rt, NULL, NULL); range_tree_destroy(rt); return (error); } for (int c = 0; c < vd->vdev_children; c++) { error = vdev_dtl_load(vd->vdev_child[c]); if (error != 0) break; } return (error); } static void vdev_zap_allocation_data(vdev_t *vd, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; vdev_alloc_bias_t alloc_bias = vd->vdev_alloc_bias; const char *string; ASSERT(alloc_bias != VDEV_BIAS_NONE); string = (alloc_bias == VDEV_BIAS_LOG) ? VDEV_ALLOC_BIAS_LOG : (alloc_bias == VDEV_BIAS_SPECIAL) ? VDEV_ALLOC_BIAS_SPECIAL : (alloc_bias == VDEV_BIAS_DEDUP) ? VDEV_ALLOC_BIAS_DEDUP : NULL; ASSERT(string != NULL); VERIFY0(zap_add(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_ALLOCATION_BIAS, 1, strlen(string) + 1, string, tx)); if (alloc_bias == VDEV_BIAS_SPECIAL || alloc_bias == VDEV_BIAS_DEDUP) { spa_activate_allocation_classes(spa, tx); } } void vdev_destroy_unlink_zap(vdev_t *vd, uint64_t zapobj, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; VERIFY0(zap_destroy(spa->spa_meta_objset, zapobj, tx)); VERIFY0(zap_remove_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, zapobj, tx)); } uint64_t vdev_create_link_zap(vdev_t *vd, dmu_tx_t *tx) { spa_t *spa = vd->vdev_spa; uint64_t zap = zap_create(spa->spa_meta_objset, DMU_OTN_ZAP_METADATA, DMU_OT_NONE, 0, tx); ASSERT(zap != 0); VERIFY0(zap_add_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps, zap, tx)); return (zap); } void vdev_construct_zaps(vdev_t *vd, dmu_tx_t *tx) { if (vd->vdev_ops != &vdev_hole_ops && vd->vdev_ops != &vdev_missing_ops && vd->vdev_ops != &vdev_root_ops && !vd->vdev_top->vdev_removing) { if (vd->vdev_ops->vdev_op_leaf && vd->vdev_leaf_zap == 0) { vd->vdev_leaf_zap = vdev_create_link_zap(vd, tx); } if (vd == vd->vdev_top && vd->vdev_top_zap == 0) { vd->vdev_top_zap = vdev_create_link_zap(vd, tx); if (vd->vdev_alloc_bias != VDEV_BIAS_NONE) vdev_zap_allocation_data(vd, tx); } } for (uint64_t i = 0; i < vd->vdev_children; i++) { vdev_construct_zaps(vd->vdev_child[i], tx); } } static void vdev_dtl_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; range_tree_t *rt = vd->vdev_dtl[DTL_MISSING]; objset_t *mos = spa->spa_meta_objset; range_tree_t *rtsync; dmu_tx_t *tx; uint64_t object = space_map_object(vd->vdev_dtl_sm); ASSERT(vdev_is_concrete(vd)); ASSERT(vd->vdev_ops->vdev_op_leaf); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); if (vd->vdev_detached || vd->vdev_top->vdev_removing) { mutex_enter(&vd->vdev_dtl_lock); space_map_free(vd->vdev_dtl_sm, tx); space_map_close(vd->vdev_dtl_sm); vd->vdev_dtl_sm = NULL; mutex_exit(&vd->vdev_dtl_lock); /* * We only destroy the leaf ZAP for detached leaves or for * removed log devices. Removed data devices handle leaf ZAP * cleanup later, once cancellation is no longer possible. */ if (vd->vdev_leaf_zap != 0 && (vd->vdev_detached || vd->vdev_top->vdev_islog)) { vdev_destroy_unlink_zap(vd, vd->vdev_leaf_zap, tx); vd->vdev_leaf_zap = 0; } dmu_tx_commit(tx); return; } if (vd->vdev_dtl_sm == NULL) { uint64_t new_object; new_object = space_map_alloc(mos, zfs_vdev_dtl_sm_blksz, tx); VERIFY3U(new_object, !=, 0); VERIFY0(space_map_open(&vd->vdev_dtl_sm, mos, new_object, 0, -1ULL, 0)); ASSERT(vd->vdev_dtl_sm != NULL); } rtsync = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0); mutex_enter(&vd->vdev_dtl_lock); range_tree_walk(rt, range_tree_add, rtsync); mutex_exit(&vd->vdev_dtl_lock); space_map_truncate(vd->vdev_dtl_sm, zfs_vdev_dtl_sm_blksz, tx); space_map_write(vd->vdev_dtl_sm, rtsync, SM_ALLOC, SM_NO_VDEVID, tx); range_tree_vacate(rtsync, NULL, NULL); range_tree_destroy(rtsync); /* * If the object for the space map has changed then dirty * the top level so that we update the config. */ if (object != space_map_object(vd->vdev_dtl_sm)) { vdev_dbgmsg(vd, "txg %llu, spa %s, DTL old object %llu, " "new object %llu", (u_longlong_t)txg, spa_name(spa), (u_longlong_t)object, (u_longlong_t)space_map_object(vd->vdev_dtl_sm)); vdev_config_dirty(vd->vdev_top); } dmu_tx_commit(tx); } /* * Determine whether the specified vdev can be offlined/detached/removed * without losing data. */ boolean_t vdev_dtl_required(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *tvd = vd->vdev_top; uint8_t cant_read = vd->vdev_cant_read; boolean_t required; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == spa->spa_root_vdev || vd == tvd) return (B_TRUE); /* * Temporarily mark the device as unreadable, and then determine * whether this results in any DTL outages in the top-level vdev. * If not, we can safely offline/detach/remove the device. */ vd->vdev_cant_read = B_TRUE; vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE); required = !vdev_dtl_empty(tvd, DTL_OUTAGE); vd->vdev_cant_read = cant_read; vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE); if (!required && zio_injection_enabled) { required = !!zio_handle_device_injection(vd, NULL, SET_ERROR(ECHILD)); } return (required); } /* * Determine if resilver is needed, and if so the txg range. */ boolean_t vdev_resilver_needed(vdev_t *vd, uint64_t *minp, uint64_t *maxp) { boolean_t needed = B_FALSE; uint64_t thismin = UINT64_MAX; uint64_t thismax = 0; if (vd->vdev_children == 0) { mutex_enter(&vd->vdev_dtl_lock); if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) && vdev_writeable(vd)) { thismin = vdev_dtl_min(vd); thismax = vdev_dtl_max(vd); needed = B_TRUE; } mutex_exit(&vd->vdev_dtl_lock); } else { for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; uint64_t cmin, cmax; if (vdev_resilver_needed(cvd, &cmin, &cmax)) { thismin = MIN(thismin, cmin); thismax = MAX(thismax, cmax); needed = B_TRUE; } } } if (needed && minp) { *minp = thismin; *maxp = thismax; } return (needed); } /* * Gets the checkpoint space map object from the vdev's ZAP. On success sm_obj * will contain either the checkpoint spacemap object or zero if none exists. * All other errors are returned to the caller. */ int vdev_checkpoint_sm_object(vdev_t *vd, uint64_t *sm_obj) { ASSERT0(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER)); if (vd->vdev_top_zap == 0) { *sm_obj = 0; return (0); } int error = zap_lookup(spa_meta_objset(vd->vdev_spa), vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, sizeof (uint64_t), 1, sm_obj); if (error == ENOENT) { *sm_obj = 0; error = 0; } return (error); } int vdev_load(vdev_t *vd) { int children = vd->vdev_children; int error = 0; taskq_t *tq = NULL; /* * It's only worthwhile to use the taskq for the root vdev, because the * slow part is metaslab_init, and that only happens for top-level * vdevs. */ if (vd->vdev_ops == &vdev_root_ops && vd->vdev_children > 0) { tq = taskq_create("vdev_load", children, minclsyspri, children, children, TASKQ_PREPOPULATE); } /* * Recursively load all children. */ for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; if (tq == NULL || vdev_uses_zvols(cvd)) { cvd->vdev_load_error = vdev_load(cvd); } else { VERIFY(taskq_dispatch(tq, vdev_load_child, cvd, TQ_SLEEP) != TASKQID_INVALID); } } if (tq != NULL) { taskq_wait(tq); taskq_destroy(tq); } for (int c = 0; c < vd->vdev_children; c++) { int error = vd->vdev_child[c]->vdev_load_error; if (error != 0) return (error); } vdev_set_deflate_ratio(vd); /* * On spa_load path, grab the allocation bias from our zap */ if (vd == vd->vdev_top && vd->vdev_top_zap != 0) { spa_t *spa = vd->vdev_spa; char bias_str[64]; error = zap_lookup(spa->spa_meta_objset, vd->vdev_top_zap, VDEV_TOP_ZAP_ALLOCATION_BIAS, 1, sizeof (bias_str), bias_str); if (error == 0) { ASSERT(vd->vdev_alloc_bias == VDEV_BIAS_NONE); vd->vdev_alloc_bias = vdev_derive_alloc_bias(bias_str); } else if (error != ENOENT) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: zap_lookup(top_zap=%llu) " "failed [error=%d]", (u_longlong_t)vd->vdev_top_zap, error); return (error); } } /* * Load any rebuild state from the top-level vdev zap. */ if (vd == vd->vdev_top && vd->vdev_top_zap != 0) { error = vdev_rebuild_load(vd); if (error && error != ENOTSUP) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: vdev_rebuild_load " "failed [error=%d]", error); return (error); } } /* * If this is a top-level vdev, initialize its metaslabs. */ if (vd == vd->vdev_top && vdev_is_concrete(vd)) { vdev_metaslab_group_create(vd); if (vd->vdev_ashift == 0 || vd->vdev_asize == 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: invalid size. ashift=%llu, " "asize=%llu", (u_longlong_t)vd->vdev_ashift, (u_longlong_t)vd->vdev_asize); return (SET_ERROR(ENXIO)); } error = vdev_metaslab_init(vd, 0); if (error != 0) { vdev_dbgmsg(vd, "vdev_load: metaslab_init failed " "[error=%d]", error); vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); return (error); } uint64_t checkpoint_sm_obj; error = vdev_checkpoint_sm_object(vd, &checkpoint_sm_obj); if (error == 0 && checkpoint_sm_obj != 0) { objset_t *mos = spa_meta_objset(vd->vdev_spa); ASSERT(vd->vdev_asize != 0); ASSERT3P(vd->vdev_checkpoint_sm, ==, NULL); error = space_map_open(&vd->vdev_checkpoint_sm, mos, checkpoint_sm_obj, 0, vd->vdev_asize, vd->vdev_ashift); if (error != 0) { vdev_dbgmsg(vd, "vdev_load: space_map_open " "failed for checkpoint spacemap (obj %llu) " "[error=%d]", (u_longlong_t)checkpoint_sm_obj, error); return (error); } ASSERT3P(vd->vdev_checkpoint_sm, !=, NULL); /* * Since the checkpoint_sm contains free entries * exclusively we can use space_map_allocated() to * indicate the cumulative checkpointed space that * has been freed. */ vd->vdev_stat.vs_checkpoint_space = -space_map_allocated(vd->vdev_checkpoint_sm); vd->vdev_spa->spa_checkpoint_info.sci_dspace += vd->vdev_stat.vs_checkpoint_space; } else if (error != 0) { vdev_dbgmsg(vd, "vdev_load: failed to retrieve " "checkpoint space map object from vdev ZAP " "[error=%d]", error); return (error); } } /* * If this is a leaf vdev, load its DTL. */ if (vd->vdev_ops->vdev_op_leaf && (error = vdev_dtl_load(vd)) != 0) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: vdev_dtl_load failed " "[error=%d]", error); return (error); } uint64_t obsolete_sm_object; error = vdev_obsolete_sm_object(vd, &obsolete_sm_object); if (error == 0 && obsolete_sm_object != 0) { objset_t *mos = vd->vdev_spa->spa_meta_objset; ASSERT(vd->vdev_asize != 0); ASSERT3P(vd->vdev_obsolete_sm, ==, NULL); if ((error = space_map_open(&vd->vdev_obsolete_sm, mos, obsolete_sm_object, 0, vd->vdev_asize, 0))) { vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); vdev_dbgmsg(vd, "vdev_load: space_map_open failed for " "obsolete spacemap (obj %llu) [error=%d]", (u_longlong_t)obsolete_sm_object, error); return (error); } } else if (error != 0) { vdev_dbgmsg(vd, "vdev_load: failed to retrieve obsolete " "space map object from vdev ZAP [error=%d]", error); return (error); } return (0); } /* * The special vdev case is used for hot spares and l2cache devices. Its * sole purpose it to set the vdev state for the associated vdev. To do this, * we make sure that we can open the underlying device, then try to read the * label, and make sure that the label is sane and that it hasn't been * repurposed to another pool. */ int vdev_validate_aux(vdev_t *vd) { nvlist_t *label; uint64_t guid, version; uint64_t state; if (!vdev_readable(vd)) return (0); if ((label = vdev_label_read_config(vd, -1ULL)) == NULL) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); return (-1); } if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_VERSION, &version) != 0 || !SPA_VERSION_IS_SUPPORTED(version) || nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0 || guid != vd->vdev_guid || nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) { vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); nvlist_free(label); return (-1); } /* * We don't actually check the pool state here. If it's in fact in * use by another pool, we update this fact on the fly when requested. */ nvlist_free(label); return (0); } static void vdev_destroy_ms_flush_data(vdev_t *vd, dmu_tx_t *tx) { objset_t *mos = spa_meta_objset(vd->vdev_spa); if (vd->vdev_top_zap == 0) return; uint64_t object = 0; int err = zap_lookup(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, sizeof (uint64_t), 1, &object); if (err == ENOENT) return; VERIFY0(err); VERIFY0(dmu_object_free(mos, object, tx)); VERIFY0(zap_remove(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, tx)); } /* * Free the objects used to store this vdev's spacemaps, and the array * that points to them. */ void vdev_destroy_spacemaps(vdev_t *vd, dmu_tx_t *tx) { if (vd->vdev_ms_array == 0) return; objset_t *mos = vd->vdev_spa->spa_meta_objset; uint64_t array_count = vd->vdev_asize >> vd->vdev_ms_shift; size_t array_bytes = array_count * sizeof (uint64_t); uint64_t *smobj_array = kmem_alloc(array_bytes, KM_SLEEP); VERIFY0(dmu_read(mos, vd->vdev_ms_array, 0, array_bytes, smobj_array, 0)); for (uint64_t i = 0; i < array_count; i++) { uint64_t smobj = smobj_array[i]; if (smobj == 0) continue; space_map_free_obj(mos, smobj, tx); } kmem_free(smobj_array, array_bytes); VERIFY0(dmu_object_free(mos, vd->vdev_ms_array, tx)); vdev_destroy_ms_flush_data(vd, tx); vd->vdev_ms_array = 0; } static void vdev_remove_empty_log(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; ASSERT(vd->vdev_islog); ASSERT(vd == vd->vdev_top); ASSERT3U(txg, ==, spa_syncing_txg(spa)); dmu_tx_t *tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg); vdev_destroy_spacemaps(vd, tx); if (vd->vdev_top_zap != 0) { vdev_destroy_unlink_zap(vd, vd->vdev_top_zap, tx); vd->vdev_top_zap = 0; } dmu_tx_commit(tx); } void vdev_sync_done(vdev_t *vd, uint64_t txg) { metaslab_t *msp; boolean_t reassess = !txg_list_empty(&vd->vdev_ms_list, TXG_CLEAN(txg)); ASSERT(vdev_is_concrete(vd)); while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg))) != NULL) metaslab_sync_done(msp, txg); if (reassess) { metaslab_sync_reassess(vd->vdev_mg); if (vd->vdev_log_mg != NULL) metaslab_sync_reassess(vd->vdev_log_mg); } } void vdev_sync(vdev_t *vd, uint64_t txg) { spa_t *spa = vd->vdev_spa; vdev_t *lvd; metaslab_t *msp; ASSERT3U(txg, ==, spa->spa_syncing_txg); dmu_tx_t *tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); if (range_tree_space(vd->vdev_obsolete_segments) > 0) { ASSERT(vd->vdev_removing || vd->vdev_ops == &vdev_indirect_ops); vdev_indirect_sync_obsolete(vd, tx); /* * If the vdev is indirect, it can't have dirty * metaslabs or DTLs. */ if (vd->vdev_ops == &vdev_indirect_ops) { ASSERT(txg_list_empty(&vd->vdev_ms_list, txg)); ASSERT(txg_list_empty(&vd->vdev_dtl_list, txg)); dmu_tx_commit(tx); return; } } ASSERT(vdev_is_concrete(vd)); if (vd->vdev_ms_array == 0 && vd->vdev_ms_shift != 0 && !vd->vdev_removing) { ASSERT(vd == vd->vdev_top); ASSERT0(vd->vdev_indirect_config.vic_mapping_object); vd->vdev_ms_array = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_OBJECT_ARRAY, 0, DMU_OT_NONE, 0, tx); ASSERT(vd->vdev_ms_array != 0); vdev_config_dirty(vd); } while ((msp = txg_list_remove(&vd->vdev_ms_list, txg)) != NULL) { metaslab_sync(msp, txg); (void) txg_list_add(&vd->vdev_ms_list, msp, TXG_CLEAN(txg)); } while ((lvd = txg_list_remove(&vd->vdev_dtl_list, txg)) != NULL) vdev_dtl_sync(lvd, txg); /* * If this is an empty log device being removed, destroy the * metadata associated with it. */ if (vd->vdev_islog && vd->vdev_stat.vs_alloc == 0 && vd->vdev_removing) vdev_remove_empty_log(vd, txg); (void) txg_list_add(&spa->spa_vdev_txg_list, vd, TXG_CLEAN(txg)); dmu_tx_commit(tx); } uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize) { return (vd->vdev_ops->vdev_op_asize(vd, psize)); } /* * Mark the given vdev faulted. A faulted vdev behaves as if the device could * not be opened, and no I/O is attempted. */ int vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd, *tvd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); tvd = vd->vdev_top; /* * If user did a 'zpool offline -f' then make the fault persist across * reboots. */ if (aux == VDEV_AUX_EXTERNAL_PERSIST) { /* * There are two kinds of forced faults: temporary and * persistent. Temporary faults go away at pool import, while * persistent faults stay set. Both types of faults can be * cleared with a zpool clear. * * We tell if a vdev is persistently faulted by looking at the * ZPOOL_CONFIG_AUX_STATE nvpair. If it's set to "external" at * import then it's a persistent fault. Otherwise, it's * temporary. We get ZPOOL_CONFIG_AUX_STATE set to "external" * by setting vd.vdev_stat.vs_aux to VDEV_AUX_EXTERNAL. This * tells vdev_config_generate() (which gets run later) to set * ZPOOL_CONFIG_AUX_STATE to "external" in the nvlist. */ vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL; vd->vdev_tmpoffline = B_FALSE; aux = VDEV_AUX_EXTERNAL; } else { vd->vdev_tmpoffline = B_TRUE; } /* * We don't directly use the aux state here, but if we do a * vdev_reopen(), we need this value to be present to remember why we * were faulted. */ vd->vdev_label_aux = aux; /* * Faulted state takes precedence over degraded. */ vd->vdev_delayed_close = B_FALSE; vd->vdev_faulted = 1ULL; vd->vdev_degraded = 0ULL; vdev_set_state(vd, B_FALSE, VDEV_STATE_FAULTED, aux); /* * If this device has the only valid copy of the data, then * back off and simply mark the vdev as degraded instead. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) { vd->vdev_degraded = 1ULL; vd->vdev_faulted = 0ULL; /* * If we reopen the device and it's not dead, only then do we * mark it degraded. */ vdev_reopen(tvd); if (vdev_readable(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); } return (spa_vdev_state_exit(spa, vd, 0)); } /* * Mark the given vdev degraded. A degraded vdev is purely an indication to the * user that something is wrong. The vdev continues to operate as normal as far * as I/O is concerned. */ int vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux) { vdev_t *vd; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); /* * If the vdev is already faulted, then don't do anything. */ if (vd->vdev_faulted || vd->vdev_degraded) return (spa_vdev_state_exit(spa, NULL, 0)); vd->vdev_degraded = 1ULL; if (!vdev_is_dead(vd)) vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux); return (spa_vdev_state_exit(spa, vd, 0)); } +int +vdev_remove_wanted(spa_t *spa, uint64_t guid) +{ + vdev_t *vd; + + spa_vdev_state_enter(spa, SCL_NONE); + + if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) + return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); + + /* + * If the vdev is already removed, then don't do anything. + */ + if (vd->vdev_removed) + return (spa_vdev_state_exit(spa, NULL, 0)); + + vd->vdev_remove_wanted = B_TRUE; + spa_async_request(spa, SPA_ASYNC_REMOVE); + + return (spa_vdev_state_exit(spa, vd, 0)); +} + + /* * Online the given vdev. * * If 'ZFS_ONLINE_UNSPARE' is set, it implies two things. First, any attached * spare device should be detached when the device finishes resilvering. * Second, the online should be treated like a 'test' online case, so no FMA * events are generated if the device fails to open. */ int vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *newstate) { vdev_t *vd, *tvd, *pvd, *rvd = spa->spa_root_vdev; boolean_t wasoffline; vdev_state_t oldstate; spa_vdev_state_enter(spa, SCL_NONE); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); wasoffline = (vd->vdev_offline || vd->vdev_tmpoffline); oldstate = vd->vdev_state; tvd = vd->vdev_top; vd->vdev_offline = B_FALSE; vd->vdev_tmpoffline = B_FALSE; vd->vdev_checkremove = !!(flags & ZFS_ONLINE_CHECKREMOVE); vd->vdev_forcefault = !!(flags & ZFS_ONLINE_FORCEFAULT); /* XXX - L2ARC 1.0 does not support expansion */ if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = !!((flags & ZFS_ONLINE_EXPAND) || spa->spa_autoexpand); vd->vdev_expansion_time = gethrestime_sec(); } vdev_reopen(tvd); vd->vdev_checkremove = vd->vdev_forcefault = B_FALSE; if (!vd->vdev_aux) { for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) pvd->vdev_expanding = B_FALSE; } if (newstate) *newstate = vd->vdev_state; if ((flags & ZFS_ONLINE_UNSPARE) && !vdev_is_dead(vd) && vd->vdev_parent && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; if ((flags & ZFS_ONLINE_EXPAND) || spa->spa_autoexpand) { /* XXX - L2ARC 1.0 does not support expansion */ if (vd->vdev_aux) return (spa_vdev_state_exit(spa, vd, ENOTSUP)); spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } /* Restart initializing if necessary */ mutex_enter(&vd->vdev_initialize_lock); if (vdev_writeable(vd) && vd->vdev_initialize_thread == NULL && vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) { (void) vdev_initialize(vd); } mutex_exit(&vd->vdev_initialize_lock); /* * Restart trimming if necessary. We do not restart trimming for cache * devices here. This is triggered by l2arc_rebuild_vdev() * asynchronously for the whole device or in l2arc_evict() as it evicts * space for upcoming writes. */ mutex_enter(&vd->vdev_trim_lock); if (vdev_writeable(vd) && !vd->vdev_isl2cache && vd->vdev_trim_thread == NULL && vd->vdev_trim_state == VDEV_TRIM_ACTIVE) { (void) vdev_trim(vd, vd->vdev_trim_rate, vd->vdev_trim_partial, vd->vdev_trim_secure); } mutex_exit(&vd->vdev_trim_lock); if (wasoffline || (oldstate < VDEV_STATE_DEGRADED && vd->vdev_state >= VDEV_STATE_DEGRADED)) spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_ONLINE); return (spa_vdev_state_exit(spa, vd, 0)); } static int vdev_offline_locked(spa_t *spa, uint64_t guid, uint64_t flags) { vdev_t *vd, *tvd; int error = 0; uint64_t generation; metaslab_group_t *mg; top: spa_vdev_state_enter(spa, SCL_ALLOC); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV))); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP))); if (vd->vdev_ops == &vdev_draid_spare_ops) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); tvd = vd->vdev_top; mg = tvd->vdev_mg; generation = spa->spa_config_generation + 1; /* * If the device isn't already offline, try to offline it. */ if (!vd->vdev_offline) { /* * If this device has the only valid copy of some data, * don't allow it to be offlined. Log devices are always * expendable. */ if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) return (spa_vdev_state_exit(spa, NULL, SET_ERROR(EBUSY))); /* * If the top-level is a slog and it has had allocations * then proceed. We check that the vdev's metaslab group * is not NULL since it's possible that we may have just * added this vdev but not yet initialized its metaslabs. */ if (tvd->vdev_islog && mg != NULL) { /* * Prevent any future allocations. */ ASSERT3P(tvd->vdev_log_mg, ==, NULL); metaslab_group_passivate(mg); (void) spa_vdev_state_exit(spa, vd, 0); error = spa_reset_logs(spa); /* * If the log device was successfully reset but has * checkpointed data, do not offline it. */ if (error == 0 && tvd->vdev_checkpoint_sm != NULL) { ASSERT3U(space_map_allocated( tvd->vdev_checkpoint_sm), !=, 0); error = ZFS_ERR_CHECKPOINT_EXISTS; } spa_vdev_state_enter(spa, SCL_ALLOC); /* * Check to see if the config has changed. */ if (error || generation != spa->spa_config_generation) { metaslab_group_activate(mg); if (error) return (spa_vdev_state_exit(spa, vd, error)); (void) spa_vdev_state_exit(spa, vd, 0); goto top; } ASSERT0(tvd->vdev_stat.vs_alloc); } /* * Offline this device and reopen its top-level vdev. * If the top-level vdev is a log device then just offline * it. Otherwise, if this action results in the top-level * vdev becoming unusable, undo it and fail the request. */ vd->vdev_offline = B_TRUE; vdev_reopen(tvd); if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_is_dead(tvd)) { vd->vdev_offline = B_FALSE; vdev_reopen(tvd); return (spa_vdev_state_exit(spa, NULL, SET_ERROR(EBUSY))); } /* * Add the device back into the metaslab rotor so that * once we online the device it's open for business. */ if (tvd->vdev_islog && mg != NULL) metaslab_group_activate(mg); } vd->vdev_tmpoffline = !!(flags & ZFS_OFFLINE_TEMPORARY); return (spa_vdev_state_exit(spa, vd, 0)); } int vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags) { int error; mutex_enter(&spa->spa_vdev_top_lock); error = vdev_offline_locked(spa, guid, flags); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Clear the error counts associated with this vdev. Unlike vdev_online() and * vdev_offline(), we assume the spa config is locked. We also clear all * children. If 'vd' is NULL, then the user wants to clear all vdevs. */ void vdev_clear(spa_t *spa, vdev_t *vd) { vdev_t *rvd = spa->spa_root_vdev; ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL); if (vd == NULL) vd = rvd; vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; vd->vdev_stat.vs_slow_ios = 0; for (int c = 0; c < vd->vdev_children; c++) vdev_clear(spa, vd->vdev_child[c]); /* * It makes no sense to "clear" an indirect vdev. */ if (!vdev_is_concrete(vd)) return; /* * If we're in the FAULTED state or have experienced failed I/O, then * clear the persistent state and attempt to reopen the device. We * also mark the vdev config dirty, so that the new faulted state is * written out to disk. */ if (vd->vdev_faulted || vd->vdev_degraded || !vdev_readable(vd) || !vdev_writeable(vd)) { /* * When reopening in response to a clear event, it may be due to * a fmadm repair request. In this case, if the device is * still broken, we want to still post the ereport again. */ vd->vdev_forcefault = B_TRUE; vd->vdev_faulted = vd->vdev_degraded = 0ULL; vd->vdev_cant_read = B_FALSE; vd->vdev_cant_write = B_FALSE; vd->vdev_stat.vs_aux = 0; vdev_reopen(vd == rvd ? rvd : vd->vdev_top); vd->vdev_forcefault = B_FALSE; if (vd != rvd && vdev_writeable(vd->vdev_top)) vdev_state_dirty(vd->vdev_top); /* If a resilver isn't required, check if vdevs can be culled */ if (vd->vdev_aux == NULL && !vdev_is_dead(vd) && !dsl_scan_resilvering(spa->spa_dsl_pool) && !dsl_scan_resilver_scheduled(spa->spa_dsl_pool)) spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_CLEAR); } /* * When clearing a FMA-diagnosed fault, we always want to * unspare the device, as we assume that the original spare was * done in response to the FMA fault. */ if (!vdev_is_dead(vd) && vd->vdev_parent != NULL && vd->vdev_parent->vdev_ops == &vdev_spare_ops && vd->vdev_parent->vdev_child[0] == vd) vd->vdev_unspare = B_TRUE; /* Clear recent error events cache (i.e. duplicate events tracking) */ zfs_ereport_clear(spa, vd); } boolean_t vdev_is_dead(vdev_t *vd) { /* * Holes and missing devices are always considered "dead". * This simplifies the code since we don't have to check for * these types of devices in the various code paths. * Instead we rely on the fact that we skip over dead devices * before issuing I/O to them. */ return (vd->vdev_state < VDEV_STATE_DEGRADED || vd->vdev_ops == &vdev_hole_ops || vd->vdev_ops == &vdev_missing_ops); } boolean_t vdev_readable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_read); } boolean_t vdev_writeable(vdev_t *vd) { return (!vdev_is_dead(vd) && !vd->vdev_cant_write && vdev_is_concrete(vd)); } boolean_t vdev_allocatable(vdev_t *vd) { uint64_t state = vd->vdev_state; /* * We currently allow allocations from vdevs which may be in the * process of reopening (i.e. VDEV_STATE_CLOSED). If the device * fails to reopen then we'll catch it later when we're holding * the proper locks. Note that we have to get the vdev state * in a local variable because although it changes atomically, * we're asking two separate questions about it. */ return (!(state < VDEV_STATE_DEGRADED && state != VDEV_STATE_CLOSED) && !vd->vdev_cant_write && vdev_is_concrete(vd) && vd->vdev_mg->mg_initialized); } boolean_t vdev_accessible(vdev_t *vd, zio_t *zio) { ASSERT(zio->io_vd == vd); if (vdev_is_dead(vd) || vd->vdev_remove_wanted) return (B_FALSE); if (zio->io_type == ZIO_TYPE_READ) return (!vd->vdev_cant_read); if (zio->io_type == ZIO_TYPE_WRITE) return (!vd->vdev_cant_write); return (B_TRUE); } static void vdev_get_child_stat(vdev_t *cvd, vdev_stat_t *vs, vdev_stat_t *cvs) { /* * Exclude the dRAID spare when aggregating to avoid double counting * the ops and bytes. These IOs are counted by the physical leaves. */ if (cvd->vdev_ops == &vdev_draid_spare_ops) return; for (int t = 0; t < VS_ZIO_TYPES; t++) { vs->vs_ops[t] += cvs->vs_ops[t]; vs->vs_bytes[t] += cvs->vs_bytes[t]; } cvs->vs_scan_removing = cvd->vdev_removing; } /* * Get extended stats */ static void vdev_get_child_stat_ex(vdev_t *cvd, vdev_stat_ex_t *vsx, vdev_stat_ex_t *cvsx) { (void) cvd; int t, b; for (t = 0; t < ZIO_TYPES; t++) { for (b = 0; b < ARRAY_SIZE(vsx->vsx_disk_histo[0]); b++) vsx->vsx_disk_histo[t][b] += cvsx->vsx_disk_histo[t][b]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_total_histo[0]); b++) { vsx->vsx_total_histo[t][b] += cvsx->vsx_total_histo[t][b]; } } for (t = 0; t < ZIO_PRIORITY_NUM_QUEUEABLE; t++) { for (b = 0; b < ARRAY_SIZE(vsx->vsx_queue_histo[0]); b++) { vsx->vsx_queue_histo[t][b] += cvsx->vsx_queue_histo[t][b]; } vsx->vsx_active_queue[t] += cvsx->vsx_active_queue[t]; vsx->vsx_pend_queue[t] += cvsx->vsx_pend_queue[t]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_ind_histo[0]); b++) vsx->vsx_ind_histo[t][b] += cvsx->vsx_ind_histo[t][b]; for (b = 0; b < ARRAY_SIZE(vsx->vsx_agg_histo[0]); b++) vsx->vsx_agg_histo[t][b] += cvsx->vsx_agg_histo[t][b]; } } boolean_t vdev_is_spacemap_addressable(vdev_t *vd) { if (spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_SPACEMAP_V2)) return (B_TRUE); /* * If double-word space map entries are not enabled we assume * 47 bits of the space map entry are dedicated to the entry's * offset (see SM_OFFSET_BITS in space_map.h). We then use that * to calculate the maximum address that can be described by a * space map entry for the given device. */ uint64_t shift = vd->vdev_ashift + SM_OFFSET_BITS; if (shift >= 63) /* detect potential overflow */ return (B_TRUE); return (vd->vdev_asize < (1ULL << shift)); } /* * Get statistics for the given vdev. */ static void vdev_get_stats_ex_impl(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx) { int t; /* * If we're getting stats on the root vdev, aggregate the I/O counts * over all top-level vdevs (i.e. the direct children of the root). */ if (!vd->vdev_ops->vdev_op_leaf) { if (vs) { memset(vs->vs_ops, 0, sizeof (vs->vs_ops)); memset(vs->vs_bytes, 0, sizeof (vs->vs_bytes)); } if (vsx) memset(vsx, 0, sizeof (*vsx)); for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; vdev_stat_t *cvs = &cvd->vdev_stat; vdev_stat_ex_t *cvsx = &cvd->vdev_stat_ex; vdev_get_stats_ex_impl(cvd, cvs, cvsx); if (vs) vdev_get_child_stat(cvd, vs, cvs); if (vsx) vdev_get_child_stat_ex(cvd, vsx, cvsx); } } else { /* * We're a leaf. Just copy our ZIO active queue stats in. The * other leaf stats are updated in vdev_stat_update(). */ if (!vsx) return; memcpy(vsx, &vd->vdev_stat_ex, sizeof (vd->vdev_stat_ex)); for (t = 0; t < ARRAY_SIZE(vd->vdev_queue.vq_class); t++) { vsx->vsx_active_queue[t] = vd->vdev_queue.vq_class[t].vqc_active; vsx->vsx_pend_queue[t] = avl_numnodes( &vd->vdev_queue.vq_class[t].vqc_queued_tree); } } } void vdev_get_stats_ex(vdev_t *vd, vdev_stat_t *vs, vdev_stat_ex_t *vsx) { vdev_t *tvd = vd->vdev_top; mutex_enter(&vd->vdev_stat_lock); if (vs) { memcpy(vs, &vd->vdev_stat, sizeof (*vs)); vs->vs_timestamp = gethrtime() - vs->vs_timestamp; vs->vs_state = vd->vdev_state; vs->vs_rsize = vdev_get_min_asize(vd); if (vd->vdev_ops->vdev_op_leaf) { vs->vs_pspace = vd->vdev_psize; vs->vs_rsize += VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE; /* * Report initializing progress. Since we don't * have the initializing locks held, this is only * an estimate (although a fairly accurate one). */ vs->vs_initialize_bytes_done = vd->vdev_initialize_bytes_done; vs->vs_initialize_bytes_est = vd->vdev_initialize_bytes_est; vs->vs_initialize_state = vd->vdev_initialize_state; vs->vs_initialize_action_time = vd->vdev_initialize_action_time; /* * Report manual TRIM progress. Since we don't have * the manual TRIM locks held, this is only an * estimate (although fairly accurate one). */ vs->vs_trim_notsup = !vd->vdev_has_trim; vs->vs_trim_bytes_done = vd->vdev_trim_bytes_done; vs->vs_trim_bytes_est = vd->vdev_trim_bytes_est; vs->vs_trim_state = vd->vdev_trim_state; vs->vs_trim_action_time = vd->vdev_trim_action_time; /* Set when there is a deferred resilver. */ vs->vs_resilver_deferred = vd->vdev_resilver_deferred; } /* * Report expandable space on top-level, non-auxiliary devices * only. The expandable space is reported in terms of metaslab * sized units since that determines how much space the pool * can expand. */ if (vd->vdev_aux == NULL && tvd != NULL) { vs->vs_esize = P2ALIGN( vd->vdev_max_asize - vd->vdev_asize, 1ULL << tvd->vdev_ms_shift); } vs->vs_configured_ashift = vd->vdev_top != NULL ? vd->vdev_top->vdev_ashift : vd->vdev_ashift; vs->vs_logical_ashift = vd->vdev_logical_ashift; if (vd->vdev_physical_ashift <= ASHIFT_MAX) vs->vs_physical_ashift = vd->vdev_physical_ashift; else vs->vs_physical_ashift = 0; /* * Report fragmentation and rebuild progress for top-level, * non-auxiliary, concrete devices. */ if (vd->vdev_aux == NULL && vd == vd->vdev_top && vdev_is_concrete(vd)) { /* * The vdev fragmentation rating doesn't take into * account the embedded slog metaslab (vdev_log_mg). * Since it's only one metaslab, it would have a tiny * impact on the overall fragmentation. */ vs->vs_fragmentation = (vd->vdev_mg != NULL) ? vd->vdev_mg->mg_fragmentation : 0; } vs->vs_noalloc = MAX(vd->vdev_noalloc, tvd ? tvd->vdev_noalloc : 0); } vdev_get_stats_ex_impl(vd, vs, vsx); mutex_exit(&vd->vdev_stat_lock); } void vdev_get_stats(vdev_t *vd, vdev_stat_t *vs) { return (vdev_get_stats_ex(vd, vs, NULL)); } void vdev_clear_stats(vdev_t *vd) { mutex_enter(&vd->vdev_stat_lock); vd->vdev_stat.vs_space = 0; vd->vdev_stat.vs_dspace = 0; vd->vdev_stat.vs_alloc = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_scan_stat_init(vdev_t *vd) { vdev_stat_t *vs = &vd->vdev_stat; for (int c = 0; c < vd->vdev_children; c++) vdev_scan_stat_init(vd->vdev_child[c]); mutex_enter(&vd->vdev_stat_lock); vs->vs_scan_processed = 0; mutex_exit(&vd->vdev_stat_lock); } void vdev_stat_update(zio_t *zio, uint64_t psize) { spa_t *spa = zio->io_spa; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd = zio->io_vd ? zio->io_vd : rvd; vdev_t *pvd; uint64_t txg = zio->io_txg; vdev_stat_t *vs = vd ? &vd->vdev_stat : NULL; vdev_stat_ex_t *vsx = vd ? &vd->vdev_stat_ex : NULL; zio_type_t type = zio->io_type; int flags = zio->io_flags; /* * If this i/o is a gang leader, it didn't do any actual work. */ if (zio->io_gang_tree) return; if (zio->io_error == 0) { /* * If this is a root i/o, don't count it -- we've already * counted the top-level vdevs, and vdev_get_stats() will * aggregate them when asked. This reduces contention on * the root vdev_stat_lock and implicitly handles blocks * that compress away to holes, for which there is no i/o. * (Holes never create vdev children, so all the counters * remain zero, which is what we want.) * * Note: this only applies to successful i/o (io_error == 0) * because unlike i/o counts, errors are not additive. * When reading a ditto block, for example, failure of * one top-level vdev does not imply a root-level error. */ if (vd == rvd) return; ASSERT(vd == zio->io_vd); if (flags & ZIO_FLAG_IO_BYPASS) return; mutex_enter(&vd->vdev_stat_lock); if (flags & ZIO_FLAG_IO_REPAIR) { /* * Repair is the result of a resilver issued by the * scan thread (spa_sync). */ if (flags & ZIO_FLAG_SCAN_THREAD) { dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan; dsl_scan_phys_t *scn_phys = &scn->scn_phys; uint64_t *processed = &scn_phys->scn_processed; if (vd->vdev_ops->vdev_op_leaf) atomic_add_64(processed, psize); vs->vs_scan_processed += psize; } /* * Repair is the result of a rebuild issued by the * rebuild thread (vdev_rebuild_thread). To avoid * double counting repaired bytes the virtual dRAID * spare vdev is excluded from the processed bytes. */ if (zio->io_priority == ZIO_PRIORITY_REBUILD) { vdev_t *tvd = vd->vdev_top; vdev_rebuild_t *vr = &tvd->vdev_rebuild_config; vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys; uint64_t *rebuilt = &vrp->vrp_bytes_rebuilt; if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops != &vdev_draid_spare_ops) { atomic_add_64(rebuilt, psize); } vs->vs_rebuild_processed += psize; } if (flags & ZIO_FLAG_SELF_HEAL) vs->vs_self_healed += psize; } /* * The bytes/ops/histograms are recorded at the leaf level and * aggregated into the higher level vdevs in vdev_get_stats(). */ if (vd->vdev_ops->vdev_op_leaf && (zio->io_priority < ZIO_PRIORITY_NUM_QUEUEABLE)) { zio_type_t vs_type = type; zio_priority_t priority = zio->io_priority; /* * TRIM ops and bytes are reported to user space as * ZIO_TYPE_IOCTL. This is done to preserve the * vdev_stat_t structure layout for user space. */ if (type == ZIO_TYPE_TRIM) vs_type = ZIO_TYPE_IOCTL; /* * Solely for the purposes of 'zpool iostat -lqrw' * reporting use the priority to categorize the IO. * Only the following are reported to user space: * * ZIO_PRIORITY_SYNC_READ, * ZIO_PRIORITY_SYNC_WRITE, * ZIO_PRIORITY_ASYNC_READ, * ZIO_PRIORITY_ASYNC_WRITE, * ZIO_PRIORITY_SCRUB, * ZIO_PRIORITY_TRIM, * ZIO_PRIORITY_REBUILD. */ if (priority == ZIO_PRIORITY_INITIALIZING) { ASSERT3U(type, ==, ZIO_TYPE_WRITE); priority = ZIO_PRIORITY_ASYNC_WRITE; } else if (priority == ZIO_PRIORITY_REMOVAL) { priority = ((type == ZIO_TYPE_WRITE) ? ZIO_PRIORITY_ASYNC_WRITE : ZIO_PRIORITY_ASYNC_READ); } vs->vs_ops[vs_type]++; vs->vs_bytes[vs_type] += psize; if (flags & ZIO_FLAG_DELEGATED) { vsx->vsx_agg_histo[priority] [RQ_HISTO(zio->io_size)]++; } else { vsx->vsx_ind_histo[priority] [RQ_HISTO(zio->io_size)]++; } if (zio->io_delta && zio->io_delay) { vsx->vsx_queue_histo[priority] [L_HISTO(zio->io_delta - zio->io_delay)]++; vsx->vsx_disk_histo[type] [L_HISTO(zio->io_delay)]++; vsx->vsx_total_histo[type] [L_HISTO(zio->io_delta)]++; } } mutex_exit(&vd->vdev_stat_lock); return; } if (flags & ZIO_FLAG_SPECULATIVE) return; /* * If this is an I/O error that is going to be retried, then ignore the * error. Otherwise, the user may interpret B_FAILFAST I/O errors as * hard errors, when in reality they can happen for any number of * innocuous reasons (bus resets, MPxIO link failure, etc). */ if (zio->io_error == EIO && !(zio->io_flags & ZIO_FLAG_IO_RETRY)) return; /* * Intent logs writes won't propagate their error to the root * I/O so don't mark these types of failures as pool-level * errors. */ if (zio->io_vd == NULL && (zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) return; if (type == ZIO_TYPE_WRITE && txg != 0 && (!(flags & ZIO_FLAG_IO_REPAIR) || (flags & ZIO_FLAG_SCAN_THREAD) || spa->spa_claiming)) { /* * This is either a normal write (not a repair), or it's * a repair induced by the scrub thread, or it's a repair * made by zil_claim() during spa_load() in the first txg. * In the normal case, we commit the DTL change in the same * txg as the block was born. In the scrub-induced repair * case, we know that scrubs run in first-pass syncing context, * so we commit the DTL change in spa_syncing_txg(spa). * In the zil_claim() case, we commit in spa_first_txg(spa). * * We currently do not make DTL entries for failed spontaneous * self-healing writes triggered by normal (non-scrubbing) * reads, because we have no transactional context in which to * do so -- and it's not clear that it'd be desirable anyway. */ if (vd->vdev_ops->vdev_op_leaf) { uint64_t commit_txg = txg; if (flags & ZIO_FLAG_SCAN_THREAD) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); ASSERT(spa_sync_pass(spa) == 1); vdev_dtl_dirty(vd, DTL_SCRUB, txg, 1); commit_txg = spa_syncing_txg(spa); } else if (spa->spa_claiming) { ASSERT(flags & ZIO_FLAG_IO_REPAIR); commit_txg = spa_first_txg(spa); } ASSERT(commit_txg >= spa_syncing_txg(spa)); if (vdev_dtl_contains(vd, DTL_MISSING, txg, 1)) return; for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent) vdev_dtl_dirty(pvd, DTL_PARTIAL, txg, 1); vdev_dirty(vd->vdev_top, VDD_DTL, vd, commit_txg); } if (vd != rvd) vdev_dtl_dirty(vd, DTL_MISSING, txg, 1); } } int64_t vdev_deflated_space(vdev_t *vd, int64_t space) { ASSERT((space & (SPA_MINBLOCKSIZE-1)) == 0); ASSERT(vd->vdev_deflate_ratio != 0 || vd->vdev_isl2cache); return ((space >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio); } /* * Update the in-core space usage stats for this vdev, its metaslab class, * and the root vdev. */ void vdev_space_update(vdev_t *vd, int64_t alloc_delta, int64_t defer_delta, int64_t space_delta) { (void) defer_delta; int64_t dspace_delta; spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; ASSERT(vd == vd->vdev_top); /* * Apply the inverse of the psize-to-asize (ie. RAID-Z) space-expansion * factor. We must calculate this here and not at the root vdev * because the root vdev's psize-to-asize is simply the max of its * children's, thus not accurate enough for us. */ dspace_delta = vdev_deflated_space(vd, space_delta); mutex_enter(&vd->vdev_stat_lock); /* ensure we won't underflow */ if (alloc_delta < 0) { ASSERT3U(vd->vdev_stat.vs_alloc, >=, -alloc_delta); } vd->vdev_stat.vs_alloc += alloc_delta; vd->vdev_stat.vs_space += space_delta; vd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&vd->vdev_stat_lock); /* every class but log contributes to root space stats */ if (vd->vdev_mg != NULL && !vd->vdev_islog) { ASSERT(!vd->vdev_isl2cache); mutex_enter(&rvd->vdev_stat_lock); rvd->vdev_stat.vs_alloc += alloc_delta; rvd->vdev_stat.vs_space += space_delta; rvd->vdev_stat.vs_dspace += dspace_delta; mutex_exit(&rvd->vdev_stat_lock); } /* Note: metaslab_class_space_update moved to metaslab_space_update */ } /* * Mark a top-level vdev's config as dirty, placing it on the dirty list * so that it will be written out next time the vdev configuration is synced. * If the root vdev is specified (vdev_top == NULL), dirty all top-level vdevs. */ void vdev_config_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_writeable(spa)); /* * If this is an aux vdev (as with l2cache and spare devices), then we * update the vdev config manually and set the sync flag. */ if (vd->vdev_aux != NULL) { spa_aux_vdev_t *sav = vd->vdev_aux; nvlist_t **aux; uint_t naux; for (c = 0; c < sav->sav_count; c++) { if (sav->sav_vdevs[c] == vd) break; } if (c == sav->sav_count) { /* * We're being removed. There's nothing more to do. */ ASSERT(sav->sav_sync == B_TRUE); return; } sav->sav_sync = B_TRUE; if (nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &aux, &naux) != 0) { VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_SPARES, &aux, &naux) == 0); } ASSERT(c < naux); /* * Setting the nvlist in the middle if the array is a little * sketchy, but it will work. */ nvlist_free(aux[c]); aux[c] = vdev_config_generate(spa, vd, B_TRUE, 0); return; } /* * The dirty list is protected by the SCL_CONFIG lock. The caller * must either hold SCL_CONFIG as writer, or must be the sync thread * (which holds SCL_CONFIG as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); if (vd == rvd) { for (c = 0; c < rvd->vdev_children; c++) vdev_config_dirty(rvd->vdev_child[c]); } else { ASSERT(vd == vd->vdev_top); if (!list_link_active(&vd->vdev_config_dirty_node) && vdev_is_concrete(vd)) { list_insert_head(&spa->spa_config_dirty_list, vd); } } } void vdev_config_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_CONFIG, RW_READER))); ASSERT(list_link_active(&vd->vdev_config_dirty_node)); list_remove(&spa->spa_config_dirty_list, vd); } /* * Mark a top-level vdev's state as dirty, so that the next pass of * spa_sync() can convert this into vdev_config_dirty(). We distinguish * the state changes from larger config changes because they require * much less locking, and are often needed for administrative actions. */ void vdev_state_dirty(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_writeable(spa)); ASSERT(vd == vd->vdev_top); /* * The state list is protected by the SCL_STATE lock. The caller * must either hold SCL_STATE as writer, or must be the sync thread * (which holds SCL_STATE as reader). There's only one sync thread, * so this is sufficient to ensure mutual exclusion. */ ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); if (!list_link_active(&vd->vdev_state_dirty_node) && vdev_is_concrete(vd)) list_insert_head(&spa->spa_state_dirty_list, vd); } void vdev_state_clean(vdev_t *vd) { spa_t *spa = vd->vdev_spa; ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) || (dsl_pool_sync_context(spa_get_dsl(spa)) && spa_config_held(spa, SCL_STATE, RW_READER))); ASSERT(list_link_active(&vd->vdev_state_dirty_node)); list_remove(&spa->spa_state_dirty_list, vd); } /* * Propagate vdev state up from children to parent. */ void vdev_propagate_state(vdev_t *vd) { spa_t *spa = vd->vdev_spa; vdev_t *rvd = spa->spa_root_vdev; int degraded = 0, faulted = 0; int corrupted = 0; vdev_t *child; if (vd->vdev_children > 0) { for (int c = 0; c < vd->vdev_children; c++) { child = vd->vdev_child[c]; /* * Don't factor holes or indirect vdevs into the * decision. */ if (!vdev_is_concrete(child)) continue; if (!vdev_readable(child) || (!vdev_writeable(child) && spa_writeable(spa))) { /* * Root special: if there is a top-level log * device, treat the root vdev as if it were * degraded. */ if (child->vdev_islog && vd == rvd) degraded++; else faulted++; } else if (child->vdev_state <= VDEV_STATE_DEGRADED) { degraded++; } if (child->vdev_stat.vs_aux == VDEV_AUX_CORRUPT_DATA) corrupted++; } vd->vdev_ops->vdev_op_state_change(vd, faulted, degraded); /* * Root special: if there is a top-level vdev that cannot be * opened due to corrupted metadata, then propagate the root * vdev's aux state as 'corrupt' rather than 'insufficient * replicas'. */ if (corrupted && vd == rvd && rvd->vdev_state == VDEV_STATE_CANT_OPEN) vdev_set_state(rvd, B_FALSE, VDEV_STATE_CANT_OPEN, VDEV_AUX_CORRUPT_DATA); } if (vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } /* * Set a vdev's state. If this is during an open, we don't update the parent * state, because we're in the process of opening children depth-first. * Otherwise, we propagate the change to the parent. * * If this routine places a device in a faulted state, an appropriate ereport is * generated. */ void vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state, vdev_aux_t aux) { uint64_t save_state; spa_t *spa = vd->vdev_spa; if (state == vd->vdev_state) { /* * Since vdev_offline() code path is already in an offline * state we can miss a statechange event to OFFLINE. Check * the previous state to catch this condition. */ if (vd->vdev_ops->vdev_op_leaf && (state == VDEV_STATE_OFFLINE) && (vd->vdev_prevstate >= VDEV_STATE_FAULTED)) { /* post an offline state change */ zfs_post_state_change(spa, vd, vd->vdev_prevstate); } vd->vdev_stat.vs_aux = aux; return; } save_state = vd->vdev_state; vd->vdev_state = state; vd->vdev_stat.vs_aux = aux; /* * If we are setting the vdev state to anything but an open state, then * always close the underlying device unless the device has requested * a delayed close (i.e. we're about to remove or fault the device). * Otherwise, we keep accessible but invalid devices open forever. * We don't call vdev_close() itself, because that implies some extra * checks (offline, etc) that we don't want here. This is limited to * leaf devices, because otherwise closing the device will affect other * children. */ if (!vd->vdev_delayed_close && vdev_is_dead(vd) && vd->vdev_ops->vdev_op_leaf) vd->vdev_ops->vdev_op_close(vd); if (vd->vdev_removed && state == VDEV_STATE_CANT_OPEN && (aux == VDEV_AUX_OPEN_FAILED || vd->vdev_checkremove)) { /* * If the previous state is set to VDEV_STATE_REMOVED, then this * device was previously marked removed and someone attempted to * reopen it. If this failed due to a nonexistent device, then * keep the device in the REMOVED state. We also let this be if * it is one of our special test online cases, which is only * attempting to online the device and shouldn't generate an FMA * fault. */ vd->vdev_state = VDEV_STATE_REMOVED; vd->vdev_stat.vs_aux = VDEV_AUX_NONE; } else if (state == VDEV_STATE_REMOVED) { vd->vdev_removed = B_TRUE; } else if (state == VDEV_STATE_CANT_OPEN) { /* * If we fail to open a vdev during an import or recovery, we * mark it as "not available", which signifies that it was * never there to begin with. Failure to open such a device * is not considered an error. */ if ((spa_load_state(spa) == SPA_LOAD_IMPORT || spa_load_state(spa) == SPA_LOAD_RECOVER) && vd->vdev_ops->vdev_op_leaf) vd->vdev_not_present = 1; /* * Post the appropriate ereport. If the 'prevstate' field is * set to something other than VDEV_STATE_UNKNOWN, it indicates * that this is part of a vdev_reopen(). In this case, we don't * want to post the ereport if the device was already in the * CANT_OPEN state beforehand. * * If the 'checkremove' flag is set, then this is an attempt to * online the device in response to an insertion event. If we * hit this case, then we have detected an insertion event for a * faulted or offline device that wasn't in the removed state. * In this scenario, we don't post an ereport because we are * about to replace the device, or attempt an online with * vdev_forcefault, which will generate the fault for us. */ if ((vd->vdev_prevstate != state || vd->vdev_forcefault) && !vd->vdev_not_present && !vd->vdev_checkremove && vd != spa->spa_root_vdev) { const char *class; switch (aux) { case VDEV_AUX_OPEN_FAILED: class = FM_EREPORT_ZFS_DEVICE_OPEN_FAILED; break; case VDEV_AUX_CORRUPT_DATA: class = FM_EREPORT_ZFS_DEVICE_CORRUPT_DATA; break; case VDEV_AUX_NO_REPLICAS: class = FM_EREPORT_ZFS_DEVICE_NO_REPLICAS; break; case VDEV_AUX_BAD_GUID_SUM: class = FM_EREPORT_ZFS_DEVICE_BAD_GUID_SUM; break; case VDEV_AUX_TOO_SMALL: class = FM_EREPORT_ZFS_DEVICE_TOO_SMALL; break; case VDEV_AUX_BAD_LABEL: class = FM_EREPORT_ZFS_DEVICE_BAD_LABEL; break; case VDEV_AUX_BAD_ASHIFT: class = FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT; break; default: class = FM_EREPORT_ZFS_DEVICE_UNKNOWN; } (void) zfs_ereport_post(class, spa, vd, NULL, NULL, save_state); } /* Erase any notion of persistent removed state */ vd->vdev_removed = B_FALSE; } else { vd->vdev_removed = B_FALSE; } /* * Notify ZED of any significant state-change on a leaf vdev. * */ if (vd->vdev_ops->vdev_op_leaf) { /* preserve original state from a vdev_reopen() */ if ((vd->vdev_prevstate != VDEV_STATE_UNKNOWN) && (vd->vdev_prevstate != vd->vdev_state) && (save_state <= VDEV_STATE_CLOSED)) save_state = vd->vdev_prevstate; /* filter out state change due to initial vdev_open */ if (save_state > VDEV_STATE_CLOSED) zfs_post_state_change(spa, vd, save_state); } if (!isopen && vd->vdev_parent) vdev_propagate_state(vd->vdev_parent); } boolean_t vdev_children_are_offline(vdev_t *vd) { ASSERT(!vd->vdev_ops->vdev_op_leaf); for (uint64_t i = 0; i < vd->vdev_children; i++) { if (vd->vdev_child[i]->vdev_state != VDEV_STATE_OFFLINE) return (B_FALSE); } return (B_TRUE); } /* * Check the vdev configuration to ensure that it's capable of supporting * a root pool. We do not support partial configuration. */ boolean_t vdev_is_bootable(vdev_t *vd) { if (!vd->vdev_ops->vdev_op_leaf) { const char *vdev_type = vd->vdev_ops->vdev_op_type; if (strcmp(vdev_type, VDEV_TYPE_MISSING) == 0) return (B_FALSE); } for (int c = 0; c < vd->vdev_children; c++) { if (!vdev_is_bootable(vd->vdev_child[c])) return (B_FALSE); } return (B_TRUE); } boolean_t vdev_is_concrete(vdev_t *vd) { vdev_ops_t *ops = vd->vdev_ops; if (ops == &vdev_indirect_ops || ops == &vdev_hole_ops || ops == &vdev_missing_ops || ops == &vdev_root_ops) { return (B_FALSE); } else { return (B_TRUE); } } /* * Determine if a log device has valid content. If the vdev was * removed or faulted in the MOS config then we know that * the content on the log device has already been written to the pool. */ boolean_t vdev_log_state_valid(vdev_t *vd) { if (vd->vdev_ops->vdev_op_leaf && !vd->vdev_faulted && !vd->vdev_removed) return (B_TRUE); for (int c = 0; c < vd->vdev_children; c++) if (vdev_log_state_valid(vd->vdev_child[c])) return (B_TRUE); return (B_FALSE); } /* * Expand a vdev if possible. */ void vdev_expand(vdev_t *vd, uint64_t txg) { ASSERT(vd->vdev_top == vd); ASSERT(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(vdev_is_concrete(vd)); vdev_set_deflate_ratio(vd); if ((vd->vdev_asize >> vd->vdev_ms_shift) > vd->vdev_ms_count && vdev_is_concrete(vd)) { vdev_metaslab_group_create(vd); VERIFY(vdev_metaslab_init(vd, txg) == 0); vdev_config_dirty(vd); } } /* * Split a vdev. */ void vdev_split(vdev_t *vd) { vdev_t *cvd, *pvd = vd->vdev_parent; vdev_remove_child(pvd, vd); vdev_compact_children(pvd); cvd = pvd->vdev_child[0]; if (pvd->vdev_children == 1) { vdev_remove_parent(cvd); cvd->vdev_splitting = B_TRUE; } vdev_propagate_state(cvd); } void vdev_deadman(vdev_t *vd, const char *tag) { for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; vdev_deadman(cvd, tag); } if (vd->vdev_ops->vdev_op_leaf) { vdev_queue_t *vq = &vd->vdev_queue; mutex_enter(&vq->vq_lock); if (avl_numnodes(&vq->vq_active_tree) > 0) { spa_t *spa = vd->vdev_spa; zio_t *fio; uint64_t delta; zfs_dbgmsg("slow vdev: %s has %lu active IOs", vd->vdev_path, avl_numnodes(&vq->vq_active_tree)); /* * Look at the head of all the pending queues, * if any I/O has been outstanding for longer than * the spa_deadman_synctime invoke the deadman logic. */ fio = avl_first(&vq->vq_active_tree); delta = gethrtime() - fio->io_timestamp; if (delta > spa_deadman_synctime(spa)) zio_deadman(fio, tag); } mutex_exit(&vq->vq_lock); } } void vdev_defer_resilver(vdev_t *vd) { ASSERT(vd->vdev_ops->vdev_op_leaf); vd->vdev_resilver_deferred = B_TRUE; vd->vdev_spa->spa_resilver_deferred = B_TRUE; } /* * Clears the resilver deferred flag on all leaf devs under vd. Returns * B_TRUE if we have devices that need to be resilvered and are available to * accept resilver I/Os. */ boolean_t vdev_clear_resilver_deferred(vdev_t *vd, dmu_tx_t *tx) { boolean_t resilver_needed = B_FALSE; spa_t *spa = vd->vdev_spa; for (int c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; resilver_needed |= vdev_clear_resilver_deferred(cvd, tx); } if (vd == spa->spa_root_vdev && spa_feature_is_active(spa, SPA_FEATURE_RESILVER_DEFER)) { spa_feature_decr(spa, SPA_FEATURE_RESILVER_DEFER, tx); vdev_config_dirty(vd); spa->spa_resilver_deferred = B_FALSE; return (resilver_needed); } if (!vdev_is_concrete(vd) || vd->vdev_aux || !vd->vdev_ops->vdev_op_leaf) return (resilver_needed); vd->vdev_resilver_deferred = B_FALSE; return (!vdev_is_dead(vd) && !vd->vdev_offline && vdev_resilver_needed(vd, NULL, NULL)); } boolean_t vdev_xlate_is_empty(range_seg64_t *rs) { return (rs->rs_start == rs->rs_end); } /* * Translate a logical range to the first contiguous physical range for the * specified vdev_t. This function is initially called with a leaf vdev and * will walk each parent vdev until it reaches a top-level vdev. Once the * top-level is reached the physical range is initialized and the recursive * function begins to unwind. As it unwinds it calls the parent's vdev * specific translation function to do the real conversion. */ void vdev_xlate(vdev_t *vd, const range_seg64_t *logical_rs, range_seg64_t *physical_rs, range_seg64_t *remain_rs) { /* * Walk up the vdev tree */ if (vd != vd->vdev_top) { vdev_xlate(vd->vdev_parent, logical_rs, physical_rs, remain_rs); } else { /* * We've reached the top-level vdev, initialize the physical * range to the logical range and set an empty remaining * range then start to unwind. */ physical_rs->rs_start = logical_rs->rs_start; physical_rs->rs_end = logical_rs->rs_end; remain_rs->rs_start = logical_rs->rs_start; remain_rs->rs_end = logical_rs->rs_start; return; } vdev_t *pvd = vd->vdev_parent; ASSERT3P(pvd, !=, NULL); ASSERT3P(pvd->vdev_ops->vdev_op_xlate, !=, NULL); /* * As this recursive function unwinds, translate the logical * range into its physical and any remaining components by calling * the vdev specific translate function. */ range_seg64_t intermediate = { 0 }; pvd->vdev_ops->vdev_op_xlate(vd, physical_rs, &intermediate, remain_rs); physical_rs->rs_start = intermediate.rs_start; physical_rs->rs_end = intermediate.rs_end; } void vdev_xlate_walk(vdev_t *vd, const range_seg64_t *logical_rs, vdev_xlate_func_t *func, void *arg) { range_seg64_t iter_rs = *logical_rs; range_seg64_t physical_rs; range_seg64_t remain_rs; while (!vdev_xlate_is_empty(&iter_rs)) { vdev_xlate(vd, &iter_rs, &physical_rs, &remain_rs); /* * With raidz and dRAID, it's possible that the logical range * does not live on this leaf vdev. Only when there is a non- * zero physical size call the provided function. */ if (!vdev_xlate_is_empty(&physical_rs)) func(arg, &physical_rs); iter_rs = remain_rs; } } static char * vdev_name(vdev_t *vd, char *buf, int buflen) { if (vd->vdev_path == NULL) { if (strcmp(vd->vdev_ops->vdev_op_type, "root") == 0) { strlcpy(buf, vd->vdev_spa->spa_name, buflen); } else if (!vd->vdev_ops->vdev_op_leaf) { snprintf(buf, buflen, "%s-%llu", vd->vdev_ops->vdev_op_type, (u_longlong_t)vd->vdev_id); } } else { strlcpy(buf, vd->vdev_path, buflen); } return (buf); } /* * Look at the vdev tree and determine whether any devices are currently being * replaced. */ boolean_t vdev_replace_in_progress(vdev_t *vdev) { ASSERT(spa_config_held(vdev->vdev_spa, SCL_ALL, RW_READER) != 0); if (vdev->vdev_ops == &vdev_replacing_ops) return (B_TRUE); /* * A 'spare' vdev indicates that we have a replace in progress, unless * it has exactly two children, and the second, the hot spare, has * finished being resilvered. */ if (vdev->vdev_ops == &vdev_spare_ops && (vdev->vdev_children > 2 || !vdev_dtl_empty(vdev->vdev_child[1], DTL_MISSING))) return (B_TRUE); for (int i = 0; i < vdev->vdev_children; i++) { if (vdev_replace_in_progress(vdev->vdev_child[i])) return (B_TRUE); } return (B_FALSE); } /* * Add a (source=src, propname=propval) list to an nvlist. */ static void vdev_prop_add_list(nvlist_t *nvl, const char *propname, char *strval, uint64_t intval, zprop_source_t src) { nvlist_t *propval; propval = fnvlist_alloc(); fnvlist_add_uint64(propval, ZPROP_SOURCE, src); if (strval != NULL) fnvlist_add_string(propval, ZPROP_VALUE, strval); else fnvlist_add_uint64(propval, ZPROP_VALUE, intval); fnvlist_add_nvlist(nvl, propname, propval); nvlist_free(propval); } static void vdev_props_set_sync(void *arg, dmu_tx_t *tx) { vdev_t *vd; nvlist_t *nvp = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; objset_t *mos = spa->spa_meta_objset; nvpair_t *elem = NULL; uint64_t vdev_guid; nvlist_t *nvprops; vdev_guid = fnvlist_lookup_uint64(nvp, ZPOOL_VDEV_PROPS_SET_VDEV); nvprops = fnvlist_lookup_nvlist(nvp, ZPOOL_VDEV_PROPS_SET_PROPS); vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE); /* this vdev could get removed while waiting for this sync task */ if (vd == NULL) return; mutex_enter(&spa->spa_props_lock); while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) { uint64_t intval, objid = 0; char *strval; vdev_prop_t prop; const char *propname = nvpair_name(elem); zprop_type_t proptype; /* * Set vdev property values in the vdev props mos object. */ if (vd->vdev_top_zap != 0) { objid = vd->vdev_top_zap; } else if (vd->vdev_leaf_zap != 0) { objid = vd->vdev_leaf_zap; } else { panic("vdev not top or leaf"); } switch (prop = vdev_name_to_prop(propname)) { case VDEV_PROP_USERPROP: if (vdev_prop_user(propname)) { strval = fnvpair_value_string(elem); if (strlen(strval) == 0) { /* remove the property if value == "" */ (void) zap_remove(mos, objid, propname, tx); } else { VERIFY0(zap_update(mos, objid, propname, 1, strlen(strval) + 1, strval, tx)); } spa_history_log_internal(spa, "vdev set", tx, "vdev_guid=%llu: %s=%s", (u_longlong_t)vdev_guid, nvpair_name(elem), strval); } break; default: /* normalize the property name */ propname = vdev_prop_to_name(prop); proptype = vdev_prop_get_type(prop); if (nvpair_type(elem) == DATA_TYPE_STRING) { ASSERT(proptype == PROP_TYPE_STRING); strval = fnvpair_value_string(elem); VERIFY0(zap_update(mos, objid, propname, 1, strlen(strval) + 1, strval, tx)); spa_history_log_internal(spa, "vdev set", tx, "vdev_guid=%llu: %s=%s", (u_longlong_t)vdev_guid, nvpair_name(elem), strval); } else if (nvpair_type(elem) == DATA_TYPE_UINT64) { intval = fnvpair_value_uint64(elem); if (proptype == PROP_TYPE_INDEX) { const char *unused; VERIFY0(vdev_prop_index_to_string( prop, intval, &unused)); } VERIFY0(zap_update(mos, objid, propname, sizeof (uint64_t), 1, &intval, tx)); spa_history_log_internal(spa, "vdev set", tx, "vdev_guid=%llu: %s=%lld", (u_longlong_t)vdev_guid, nvpair_name(elem), (longlong_t)intval); } else { panic("invalid vdev property type %u", nvpair_type(elem)); } } } mutex_exit(&spa->spa_props_lock); } int vdev_prop_set(vdev_t *vd, nvlist_t *innvl, nvlist_t *outnvl) { spa_t *spa = vd->vdev_spa; nvpair_t *elem = NULL; uint64_t vdev_guid; nvlist_t *nvprops; int error; ASSERT(vd != NULL); if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_SET_VDEV, &vdev_guid) != 0) return (SET_ERROR(EINVAL)); if (nvlist_lookup_nvlist(innvl, ZPOOL_VDEV_PROPS_SET_PROPS, &nvprops) != 0) return (SET_ERROR(EINVAL)); if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL) return (SET_ERROR(EINVAL)); while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) { char *propname = nvpair_name(elem); vdev_prop_t prop = vdev_name_to_prop(propname); uint64_t intval = 0; char *strval = NULL; if (prop == VDEV_PROP_USERPROP && !vdev_prop_user(propname)) { error = EINVAL; goto end; } if (vdev_prop_readonly(prop)) { error = EROFS; goto end; } /* Special Processing */ switch (prop) { case VDEV_PROP_PATH: if (vd->vdev_path == NULL) { error = EROFS; break; } if (nvpair_value_string(elem, &strval) != 0) { error = EINVAL; break; } /* New path must start with /dev/ */ if (strncmp(strval, "/dev/", 5)) { error = EINVAL; break; } error = spa_vdev_setpath(spa, vdev_guid, strval); break; case VDEV_PROP_ALLOCATING: if (nvpair_value_uint64(elem, &intval) != 0) { error = EINVAL; break; } if (intval != vd->vdev_noalloc) break; if (intval == 0) error = spa_vdev_noalloc(spa, vdev_guid); else error = spa_vdev_alloc(spa, vdev_guid); break; default: /* Most processing is done in vdev_props_set_sync */ break; } end: if (error != 0) { intval = error; vdev_prop_add_list(outnvl, propname, strval, intval, 0); return (error); } } return (dsl_sync_task(spa->spa_name, NULL, vdev_props_set_sync, innvl, 6, ZFS_SPACE_CHECK_EXTRA_RESERVED)); } int vdev_prop_get(vdev_t *vd, nvlist_t *innvl, nvlist_t *outnvl) { spa_t *spa = vd->vdev_spa; objset_t *mos = spa->spa_meta_objset; int err = 0; uint64_t objid; uint64_t vdev_guid; nvpair_t *elem = NULL; nvlist_t *nvprops = NULL; uint64_t intval = 0; char *strval = NULL; const char *propname = NULL; vdev_prop_t prop; ASSERT(vd != NULL); ASSERT(mos != NULL); if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_GET_VDEV, &vdev_guid) != 0) return (SET_ERROR(EINVAL)); nvlist_lookup_nvlist(innvl, ZPOOL_VDEV_PROPS_GET_PROPS, &nvprops); if (vd->vdev_top_zap != 0) { objid = vd->vdev_top_zap; } else if (vd->vdev_leaf_zap != 0) { objid = vd->vdev_leaf_zap; } else { return (SET_ERROR(EINVAL)); } ASSERT(objid != 0); mutex_enter(&spa->spa_props_lock); if (nvprops != NULL) { char namebuf[64] = { 0 }; while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) { intval = 0; strval = NULL; propname = nvpair_name(elem); prop = vdev_name_to_prop(propname); zprop_source_t src = ZPROP_SRC_DEFAULT; uint64_t integer_size, num_integers; switch (prop) { /* Special Read-only Properties */ case VDEV_PROP_NAME: strval = vdev_name(vd, namebuf, sizeof (namebuf)); if (strval == NULL) continue; vdev_prop_add_list(outnvl, propname, strval, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_CAPACITY: /* percent used */ intval = (vd->vdev_stat.vs_dspace == 0) ? 0 : (vd->vdev_stat.vs_alloc * 100 / vd->vdev_stat.vs_dspace); vdev_prop_add_list(outnvl, propname, NULL, intval, ZPROP_SRC_NONE); continue; case VDEV_PROP_STATE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_state, ZPROP_SRC_NONE); continue; case VDEV_PROP_GUID: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_guid, ZPROP_SRC_NONE); continue; case VDEV_PROP_ASIZE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_asize, ZPROP_SRC_NONE); continue; case VDEV_PROP_PSIZE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_psize, ZPROP_SRC_NONE); continue; case VDEV_PROP_ASHIFT: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_ashift, ZPROP_SRC_NONE); continue; case VDEV_PROP_SIZE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_dspace, ZPROP_SRC_NONE); continue; case VDEV_PROP_FREE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_dspace - vd->vdev_stat.vs_alloc, ZPROP_SRC_NONE); continue; case VDEV_PROP_ALLOCATED: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_alloc, ZPROP_SRC_NONE); continue; case VDEV_PROP_EXPANDSZ: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_esize, ZPROP_SRC_NONE); continue; case VDEV_PROP_FRAGMENTATION: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_fragmentation, ZPROP_SRC_NONE); continue; case VDEV_PROP_PARITY: vdev_prop_add_list(outnvl, propname, NULL, vdev_get_nparity(vd), ZPROP_SRC_NONE); continue; case VDEV_PROP_PATH: if (vd->vdev_path == NULL) continue; vdev_prop_add_list(outnvl, propname, vd->vdev_path, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_DEVID: if (vd->vdev_devid == NULL) continue; vdev_prop_add_list(outnvl, propname, vd->vdev_devid, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_PHYS_PATH: if (vd->vdev_physpath == NULL) continue; vdev_prop_add_list(outnvl, propname, vd->vdev_physpath, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_ENC_PATH: if (vd->vdev_enc_sysfs_path == NULL) continue; vdev_prop_add_list(outnvl, propname, vd->vdev_enc_sysfs_path, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_FRU: if (vd->vdev_fru == NULL) continue; vdev_prop_add_list(outnvl, propname, vd->vdev_fru, 0, ZPROP_SRC_NONE); continue; case VDEV_PROP_PARENT: if (vd->vdev_parent != NULL) { strval = vdev_name(vd->vdev_parent, namebuf, sizeof (namebuf)); vdev_prop_add_list(outnvl, propname, strval, 0, ZPROP_SRC_NONE); } continue; case VDEV_PROP_CHILDREN: if (vd->vdev_children > 0) strval = kmem_zalloc(ZAP_MAXVALUELEN, KM_SLEEP); for (uint64_t i = 0; i < vd->vdev_children; i++) { const char *vname; vname = vdev_name(vd->vdev_child[i], namebuf, sizeof (namebuf)); if (vname == NULL) vname = "(unknown)"; if (strlen(strval) > 0) strlcat(strval, ",", ZAP_MAXVALUELEN); strlcat(strval, vname, ZAP_MAXVALUELEN); } if (strval != NULL) { vdev_prop_add_list(outnvl, propname, strval, 0, ZPROP_SRC_NONE); kmem_free(strval, ZAP_MAXVALUELEN); } continue; case VDEV_PROP_NUMCHILDREN: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_children, ZPROP_SRC_NONE); continue; case VDEV_PROP_READ_ERRORS: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_read_errors, ZPROP_SRC_NONE); continue; case VDEV_PROP_WRITE_ERRORS: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_write_errors, ZPROP_SRC_NONE); continue; case VDEV_PROP_CHECKSUM_ERRORS: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_checksum_errors, ZPROP_SRC_NONE); continue; case VDEV_PROP_INITIALIZE_ERRORS: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_initialize_errors, ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_NULL: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_NULL], ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_READ: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_READ], ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_WRITE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_WRITE], ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_FREE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_FREE], ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_CLAIM: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_CLAIM], ZPROP_SRC_NONE); continue; case VDEV_PROP_OPS_TRIM: /* * TRIM ops and bytes are reported to user * space as ZIO_TYPE_IOCTL. This is done to * preserve the vdev_stat_t structure layout * for user space. */ vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_ops[ZIO_TYPE_IOCTL], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_NULL: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_NULL], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_READ: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_READ], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_WRITE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_WRITE], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_FREE: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_FREE], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_CLAIM: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_CLAIM], ZPROP_SRC_NONE); continue; case VDEV_PROP_BYTES_TRIM: /* * TRIM ops and bytes are reported to user * space as ZIO_TYPE_IOCTL. This is done to * preserve the vdev_stat_t structure layout * for user space. */ vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_stat.vs_bytes[ZIO_TYPE_IOCTL], ZPROP_SRC_NONE); continue; case VDEV_PROP_REMOVING: vdev_prop_add_list(outnvl, propname, NULL, vd->vdev_removing, ZPROP_SRC_NONE); continue; /* Numeric Properites */ case VDEV_PROP_ALLOCATING: src = ZPROP_SRC_LOCAL; strval = NULL; err = zap_lookup(mos, objid, nvpair_name(elem), sizeof (uint64_t), 1, &intval); if (err == ENOENT) { intval = vdev_prop_default_numeric(prop); err = 0; } else if (err) break; if (intval == vdev_prop_default_numeric(prop)) src = ZPROP_SRC_DEFAULT; /* Leaf vdevs cannot have this property */ if (vd->vdev_mg == NULL && vd->vdev_top != NULL) { src = ZPROP_SRC_NONE; intval = ZPROP_BOOLEAN_NA; } vdev_prop_add_list(outnvl, propname, strval, intval, src); break; /* Text Properties */ case VDEV_PROP_COMMENT: /* Exists in the ZAP below */ /* FALLTHRU */ case VDEV_PROP_USERPROP: /* User Properites */ src = ZPROP_SRC_LOCAL; err = zap_length(mos, objid, nvpair_name(elem), &integer_size, &num_integers); if (err) break; switch (integer_size) { case 8: /* User properties cannot be integers */ err = EINVAL; break; case 1: /* string property */ strval = kmem_alloc(num_integers, KM_SLEEP); err = zap_lookup(mos, objid, nvpair_name(elem), 1, num_integers, strval); if (err) { kmem_free(strval, num_integers); break; } vdev_prop_add_list(outnvl, propname, strval, 0, src); kmem_free(strval, num_integers); break; } break; default: err = ENOENT; break; } if (err) break; } } else { /* * Get all properties from the MOS vdev property object. */ zap_cursor_t zc; zap_attribute_t za; for (zap_cursor_init(&zc, mos, objid); (err = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { intval = 0; strval = NULL; zprop_source_t src = ZPROP_SRC_DEFAULT; propname = za.za_name; prop = vdev_name_to_prop(propname); switch (za.za_integer_length) { case 8: /* We do not allow integer user properties */ /* This is likely an internal value */ break; case 1: /* string property */ strval = kmem_alloc(za.za_num_integers, KM_SLEEP); err = zap_lookup(mos, objid, za.za_name, 1, za.za_num_integers, strval); if (err) { kmem_free(strval, za.za_num_integers); break; } vdev_prop_add_list(outnvl, propname, strval, 0, src); kmem_free(strval, za.za_num_integers); break; default: break; } } zap_cursor_fini(&zc); } mutex_exit(&spa->spa_props_lock); if (err && err != ENOENT) { return (err); } return (0); } EXPORT_SYMBOL(vdev_fault); EXPORT_SYMBOL(vdev_degrade); EXPORT_SYMBOL(vdev_online); EXPORT_SYMBOL(vdev_offline); EXPORT_SYMBOL(vdev_clear); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_count, UINT, ZMOD_RW, "Target number of metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_shift, UINT, ZMOD_RW, "Default limit for metaslab size"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, min_ms_count, UINT, ZMOD_RW, "Minimum number of metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, ms_count_limit, UINT, ZMOD_RW, "Practical upper limit of total metaslabs per top-level vdev"); ZFS_MODULE_PARAM(zfs, zfs_, slow_io_events_per_second, UINT, ZMOD_RW, "Rate limit slow IO (delay) events to this many per second"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM(zfs, zfs_, checksum_events_per_second, UINT, ZMOD_RW, "Rate limit checksum events to this many checksum errors per second " "(do not set below ZED threshold)."); /* END CSTYLED */ ZFS_MODULE_PARAM(zfs, zfs_, scan_ignore_errors, INT, ZMOD_RW, "Ignore errors during resilver/scrub"); ZFS_MODULE_PARAM(zfs_vdev, vdev_, validate_skip, INT, ZMOD_RW, "Bypass vdev_validate()"); ZFS_MODULE_PARAM(zfs, zfs_, nocacheflush, INT, ZMOD_RW, "Disable cache flushes"); ZFS_MODULE_PARAM(zfs, zfs_, embedded_slog_min_ms, UINT, ZMOD_RW, "Minimum number of metaslabs required to dedicate one for log blocks"); /* BEGIN CSTYLED */ ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, min_auto_ashift, param_set_min_auto_ashift, param_get_ulong, ZMOD_RW, "Minimum ashift used when creating new top-level vdevs"); ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, max_auto_ashift, param_set_max_auto_ashift, param_get_ulong, ZMOD_RW, "Maximum ashift used when optimizing for logical -> physical sector " "size on new top-level vdevs"); /* END CSTYLED */ diff --git a/module/zfs/zfs_ioctl.c b/module/zfs/zfs_ioctl.c index 620238b7265f..c3266c09306b 100644 --- a/module/zfs/zfs_ioctl.c +++ b/module/zfs/zfs_ioctl.c @@ -1,7887 +1,7891 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Portions Copyright 2011 Martin Matuska * Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved. * Portions Copyright 2012 Pawel Jakub Dawidek * Copyright (c) 2014, 2016 Joyent, Inc. All rights reserved. * Copyright 2016 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014, Joyent, Inc. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Toomas Soome * Copyright (c) 2016 Actifio, Inc. All rights reserved. * Copyright (c) 2018, loli10K . All rights reserved. * Copyright 2017 RackTop Systems. * Copyright (c) 2017 Open-E, Inc. All Rights Reserved. * Copyright (c) 2019 Datto Inc. * Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved. * Copyright (c) 2019, 2021, Klara Inc. * Copyright (c) 2019, Allan Jude */ /* * ZFS ioctls. * * This file handles the ioctls to /dev/zfs, used for configuring ZFS storage * pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool. * * There are two ways that we handle ioctls: the legacy way where almost * all of the logic is in the ioctl callback, and the new way where most * of the marshalling is handled in the common entry point, zfsdev_ioctl(). * * Non-legacy ioctls should be registered by calling * zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked * from userland by lzc_ioctl(). * * The registration arguments are as follows: * * const char *name * The name of the ioctl. This is used for history logging. If the * ioctl returns successfully (the callback returns 0), and allow_log * is true, then a history log entry will be recorded with the input & * output nvlists. The log entry can be printed with "zpool history -i". * * zfs_ioc_t ioc * The ioctl request number, which userland will pass to ioctl(2). * We want newer versions of libzfs and libzfs_core to run against * existing zfs kernel modules (i.e. a deferred reboot after an update). * Therefore the ioctl numbers cannot change from release to release. * * zfs_secpolicy_func_t *secpolicy * This function will be called before the zfs_ioc_func_t, to * determine if this operation is permitted. It should return EPERM * on failure, and 0 on success. Checks include determining if the * dataset is visible in this zone, and if the user has either all * zfs privileges in the zone (SYS_MOUNT), or has been granted permission * to do this operation on this dataset with "zfs allow". * * zfs_ioc_namecheck_t namecheck * This specifies what to expect in the zfs_cmd_t:zc_name -- a pool * name, a dataset name, or nothing. If the name is not well-formed, * the ioctl will fail and the callback will not be called. * Therefore, the callback can assume that the name is well-formed * (e.g. is null-terminated, doesn't have more than one '@' character, * doesn't have invalid characters). * * zfs_ioc_poolcheck_t pool_check * This specifies requirements on the pool state. If the pool does * not meet them (is suspended or is readonly), the ioctl will fail * and the callback will not be called. If any checks are specified * (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME. * Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED | * POOL_CHECK_READONLY). * * zfs_ioc_key_t *nvl_keys * The list of expected/allowable innvl input keys. This list is used * to validate the nvlist input to the ioctl. * * boolean_t smush_outnvlist * If smush_outnvlist is true, then the output is presumed to be a * list of errors, and it will be "smushed" down to fit into the * caller's buffer, by removing some entries and replacing them with a * single "N_MORE_ERRORS" entry indicating how many were removed. See * nvlist_smush() for details. If smush_outnvlist is false, and the * outnvlist does not fit into the userland-provided buffer, then the * ioctl will fail with ENOMEM. * * zfs_ioc_func_t *func * The callback function that will perform the operation. * * The callback should return 0 on success, or an error number on * failure. If the function fails, the userland ioctl will return -1, * and errno will be set to the callback's return value. The callback * will be called with the following arguments: * * const char *name * The name of the pool or dataset to operate on, from * zfs_cmd_t:zc_name. The 'namecheck' argument specifies the * expected type (pool, dataset, or none). * * nvlist_t *innvl * The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or * NULL if no input nvlist was provided. Changes to this nvlist are * ignored. If the input nvlist could not be deserialized, the * ioctl will fail and the callback will not be called. * * nvlist_t *outnvl * The output nvlist, initially empty. The callback can fill it in, * and it will be returned to userland by serializing it into * zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization * fails (e.g. because the caller didn't supply a large enough * buffer), then the overall ioctl will fail. See the * 'smush_nvlist' argument above for additional behaviors. * * There are two typical uses of the output nvlist: * - To return state, e.g. property values. In this case, * smush_outnvlist should be false. If the buffer was not large * enough, the caller will reallocate a larger buffer and try * the ioctl again. * * - To return multiple errors from an ioctl which makes on-disk * changes. In this case, smush_outnvlist should be true. * Ioctls which make on-disk modifications should generally not * use the outnvl if they succeed, because the caller can not * distinguish between the operation failing, and * deserialization failing. * * IOCTL Interface Errors * * The following ioctl input errors can be returned: * ZFS_ERR_IOC_CMD_UNAVAIL the ioctl number is not supported by kernel * ZFS_ERR_IOC_ARG_UNAVAIL an input argument is not supported by kernel * ZFS_ERR_IOC_ARG_REQUIRED a required input argument is missing * ZFS_ERR_IOC_ARG_BADTYPE an input argument has an invalid type */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_namecheck.h" #include "zfs_prop.h" #include "zfs_deleg.h" #include "zfs_comutil.h" #include #include #include kmutex_t zfsdev_state_lock; static zfsdev_state_t *zfsdev_state_list; /* * Limit maximum nvlist size. We don't want users passing in insane values * for zc->zc_nvlist_src_size, since we will need to allocate that much memory. * Defaults to 0=auto which is handled by platform code. */ unsigned long zfs_max_nvlist_src_size = 0; /* * When logging the output nvlist of an ioctl in the on-disk history, limit * the logged size to this many bytes. This must be less than DMU_MAX_ACCESS. * This applies primarily to zfs_ioc_channel_program(). */ static unsigned long zfs_history_output_max = 1024 * 1024; uint_t zfs_fsyncer_key; uint_t zfs_allow_log_key; /* DATA_TYPE_ANY is used when zkey_type can vary. */ #define DATA_TYPE_ANY DATA_TYPE_UNKNOWN typedef struct zfs_ioc_vec { zfs_ioc_legacy_func_t *zvec_legacy_func; zfs_ioc_func_t *zvec_func; zfs_secpolicy_func_t *zvec_secpolicy; zfs_ioc_namecheck_t zvec_namecheck; boolean_t zvec_allow_log; zfs_ioc_poolcheck_t zvec_pool_check; boolean_t zvec_smush_outnvlist; const char *zvec_name; const zfs_ioc_key_t *zvec_nvl_keys; size_t zvec_nvl_key_count; } zfs_ioc_vec_t; /* This array is indexed by zfs_userquota_prop_t */ static const char *userquota_perms[] = { ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_PERM_GROUPUSED, ZFS_DELEG_PERM_GROUPQUOTA, ZFS_DELEG_PERM_USEROBJUSED, ZFS_DELEG_PERM_USEROBJQUOTA, ZFS_DELEG_PERM_GROUPOBJUSED, ZFS_DELEG_PERM_GROUPOBJQUOTA, ZFS_DELEG_PERM_PROJECTUSED, ZFS_DELEG_PERM_PROJECTQUOTA, ZFS_DELEG_PERM_PROJECTOBJUSED, ZFS_DELEG_PERM_PROJECTOBJQUOTA, }; static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc); static int zfs_ioc_id_quota_upgrade(zfs_cmd_t *zc); static int zfs_check_settable(const char *name, nvpair_t *property, cred_t *cr); static int zfs_check_clearable(const char *dataset, nvlist_t *props, nvlist_t **errors); static int zfs_fill_zplprops_root(uint64_t, nvlist_t *, nvlist_t *, boolean_t *); int zfs_set_prop_nvlist(const char *, zprop_source_t, nvlist_t *, nvlist_t *); static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp); static void history_str_free(char *buf) { kmem_free(buf, HIS_MAX_RECORD_LEN); } static char * history_str_get(zfs_cmd_t *zc) { char *buf; if (zc->zc_history == 0) return (NULL); buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP); if (copyinstr((void *)(uintptr_t)zc->zc_history, buf, HIS_MAX_RECORD_LEN, NULL) != 0) { history_str_free(buf); return (NULL); } buf[HIS_MAX_RECORD_LEN -1] = '\0'; return (buf); } /* * Return non-zero if the spa version is less than requested version. */ static int zfs_earlier_version(const char *name, int version) { spa_t *spa; if (spa_open(name, &spa, FTAG) == 0) { if (spa_version(spa) < version) { spa_close(spa, FTAG); return (1); } spa_close(spa, FTAG); } return (0); } /* * Return TRUE if the ZPL version is less than requested version. */ static boolean_t zpl_earlier_version(const char *name, int version) { objset_t *os; boolean_t rc = B_TRUE; if (dmu_objset_hold(name, FTAG, &os) == 0) { uint64_t zplversion; if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (B_TRUE); } /* XXX reading from non-owned objset */ if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &zplversion) == 0) rc = zplversion < version; dmu_objset_rele(os, FTAG); } return (rc); } static void zfs_log_history(zfs_cmd_t *zc) { spa_t *spa; char *buf; if ((buf = history_str_get(zc)) == NULL) return; if (spa_open(zc->zc_name, &spa, FTAG) == 0) { if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY) (void) spa_history_log(spa, buf); spa_close(spa, FTAG); } history_str_free(buf); } /* * Policy for top-level read operations (list pools). Requires no privileges, * and can be used in the local zone, as there is no associated dataset. */ static int zfs_secpolicy_none(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl, (void) cr; return (0); } /* * Policy for dataset read operations (list children, get statistics). Requires * no privileges, but must be visible in the local zone. */ static int zfs_secpolicy_read(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl, (void) cr; if (INGLOBALZONE(curproc) || zone_dataset_visible(zc->zc_name, NULL)) return (0); return (SET_ERROR(ENOENT)); } static int zfs_dozonecheck_impl(const char *dataset, uint64_t zoned, cred_t *cr) { int writable = 1; /* * The dataset must be visible by this zone -- check this first * so they don't see EPERM on something they shouldn't know about. */ if (!INGLOBALZONE(curproc) && !zone_dataset_visible(dataset, &writable)) return (SET_ERROR(ENOENT)); if (INGLOBALZONE(curproc)) { /* * If the fs is zoned, only root can access it from the * global zone. */ if (secpolicy_zfs(cr) && zoned) return (SET_ERROR(EPERM)); } else { /* * If we are in a local zone, the 'zoned' property must be set. */ if (!zoned) return (SET_ERROR(EPERM)); /* must be writable by this zone */ if (!writable) return (SET_ERROR(EPERM)); } return (0); } static int zfs_dozonecheck(const char *dataset, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_integer(dataset, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_dozonecheck_ds(const char *dataset, dsl_dataset_t *ds, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_int_ds(ds, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_secpolicy_write_perms_ds(const char *name, dsl_dataset_t *ds, const char *perm, cred_t *cr) { int error; error = zfs_dozonecheck_ds(name, ds, cr); if (error == 0) { error = secpolicy_zfs(cr); if (error != 0) error = dsl_deleg_access_impl(ds, perm, cr); } return (error); } static int zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr) { int error; dsl_dataset_t *ds; dsl_pool_t *dp; /* * First do a quick check for root in the global zone, which * is allowed to do all write_perms. This ensures that zfs_ioc_* * will get to handle nonexistent datasets. */ if (INGLOBALZONE(curproc) && secpolicy_zfs(cr) == 0) return (0); error = dsl_pool_hold(name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(name, ds, perm, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* * Policy for setting the security label property. * * Returns 0 for success, non-zero for access and other errors. */ static int zfs_set_slabel_policy(const char *name, const char *strval, cred_t *cr) { #ifdef HAVE_MLSLABEL char ds_hexsl[MAXNAMELEN]; bslabel_t ds_sl, new_sl; boolean_t new_default = FALSE; uint64_t zoned; int needed_priv = -1; int error; /* First get the existing dataset label. */ error = dsl_prop_get(name, zfs_prop_to_name(ZFS_PROP_MLSLABEL), 1, sizeof (ds_hexsl), &ds_hexsl, NULL); if (error != 0) return (SET_ERROR(EPERM)); if (strcasecmp(strval, ZFS_MLSLABEL_DEFAULT) == 0) new_default = TRUE; /* The label must be translatable */ if (!new_default && (hexstr_to_label(strval, &new_sl) != 0)) return (SET_ERROR(EINVAL)); /* * In a non-global zone, disallow attempts to set a label that * doesn't match that of the zone; otherwise no other checks * are needed. */ if (!INGLOBALZONE(curproc)) { if (new_default || !blequal(&new_sl, CR_SL(CRED()))) return (SET_ERROR(EPERM)); return (0); } /* * For global-zone datasets (i.e., those whose zoned property is * "off", verify that the specified new label is valid for the * global zone. */ if (dsl_prop_get_integer(name, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL)) return (SET_ERROR(EPERM)); if (!zoned) { if (zfs_check_global_label(name, strval) != 0) return (SET_ERROR(EPERM)); } /* * If the existing dataset label is nondefault, check if the * dataset is mounted (label cannot be changed while mounted). * Get the zfsvfs_t; if there isn't one, then the dataset isn't * mounted (or isn't a dataset, doesn't exist, ...). */ if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) != 0) { objset_t *os; static const char *setsl_tag = "setsl_tag"; /* * Try to own the dataset; abort if there is any error, * (e.g., already mounted, in use, or other error). */ error = dmu_objset_own(name, DMU_OST_ZFS, B_TRUE, B_TRUE, setsl_tag, &os); if (error != 0) return (SET_ERROR(EPERM)); dmu_objset_disown(os, B_TRUE, setsl_tag); if (new_default) { needed_priv = PRIV_FILE_DOWNGRADE_SL; goto out_check; } if (hexstr_to_label(strval, &new_sl) != 0) return (SET_ERROR(EPERM)); if (blstrictdom(&ds_sl, &new_sl)) needed_priv = PRIV_FILE_DOWNGRADE_SL; else if (blstrictdom(&new_sl, &ds_sl)) needed_priv = PRIV_FILE_UPGRADE_SL; } else { /* dataset currently has a default label */ if (!new_default) needed_priv = PRIV_FILE_UPGRADE_SL; } out_check: if (needed_priv != -1) return (PRIV_POLICY(cr, needed_priv, B_FALSE, EPERM, NULL)); return (0); #else return (SET_ERROR(ENOTSUP)); #endif /* HAVE_MLSLABEL */ } static int zfs_secpolicy_setprop(const char *dsname, zfs_prop_t prop, nvpair_t *propval, cred_t *cr) { char *strval; /* * Check permissions for special properties. */ switch (prop) { default: break; case ZFS_PROP_ZONED: /* * Disallow setting of 'zoned' from within a local zone. */ if (!INGLOBALZONE(curproc)) return (SET_ERROR(EPERM)); break; case ZFS_PROP_QUOTA: case ZFS_PROP_FILESYSTEM_LIMIT: case ZFS_PROP_SNAPSHOT_LIMIT: if (!INGLOBALZONE(curproc)) { uint64_t zoned; char setpoint[ZFS_MAX_DATASET_NAME_LEN]; /* * Unprivileged users are allowed to modify the * limit on things *under* (ie. contained by) * the thing they own. */ if (dsl_prop_get_integer(dsname, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, setpoint)) return (SET_ERROR(EPERM)); if (!zoned || strlen(dsname) <= strlen(setpoint)) return (SET_ERROR(EPERM)); } break; case ZFS_PROP_MLSLABEL: if (!is_system_labeled()) return (SET_ERROR(EPERM)); if (nvpair_value_string(propval, &strval) == 0) { int err; err = zfs_set_slabel_policy(dsname, strval, CRED()); if (err != 0) return (err); } break; } return (zfs_secpolicy_write_perms(dsname, zfs_prop_to_name(prop), cr)); } static int zfs_secpolicy_set_fsacl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * permission to set permissions will be evaluated later in * dsl_deleg_can_allow() */ (void) innvl; return (zfs_dozonecheck(zc->zc_name, cr)); } static int zfs_secpolicy_rollback(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_ROLLBACK, cr)); } static int zfs_secpolicy_send(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; dsl_pool_t *dp; dsl_dataset_t *ds; const char *cp; int error; /* * Generate the current snapshot name from the given objsetid, then * use that name for the secpolicy/zone checks. */ cp = strchr(zc->zc_name, '@'); if (cp == NULL) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } dsl_dataset_name(ds, zc->zc_name); error = zfs_secpolicy_write_perms_ds(zc->zc_name, ds, ZFS_DELEG_PERM_SEND, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static int zfs_secpolicy_send_new(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_SEND, cr)); } static int zfs_secpolicy_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl, (void) cr; return (SET_ERROR(ENOTSUP)); } static int zfs_secpolicy_smb_acl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl, (void) cr; return (SET_ERROR(ENOTSUP)); } static int zfs_get_parent(const char *datasetname, char *parent, int parentsize) { char *cp; /* * Remove the @bla or /bla from the end of the name to get the parent. */ (void) strlcpy(parent, datasetname, parentsize); cp = strrchr(parent, '@'); if (cp != NULL) { cp[0] = '\0'; } else { cp = strrchr(parent, '/'); if (cp == NULL) return (SET_ERROR(ENOENT)); cp[0] = '\0'; } return (0); } int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr) { int error; if ((error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr)); } static int zfs_secpolicy_destroy(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; return (zfs_secpolicy_destroy_perms(zc->zc_name, cr)); } /* * Destroying snapshots with delegated permissions requires * descendant mount and destroy permissions. */ static int zfs_secpolicy_destroy_snaps(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; nvlist_t *snaps; nvpair_t *pair, *nextpair; int error = 0; snaps = fnvlist_lookup_nvlist(innvl, "snaps"); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nextpair) { nextpair = nvlist_next_nvpair(snaps, pair); error = zfs_secpolicy_destroy_perms(nvpair_name(pair), cr); if (error == ENOENT) { /* * Ignore any snapshots that don't exist (we consider * them "already destroyed"). Remove the name from the * nvl here in case the snapshot is created between * now and when we try to destroy it (in which case * we don't want to destroy it since we haven't * checked for permission). */ fnvlist_remove_nvpair(snaps, pair); error = 0; } if (error != 0) break; } return (error); } int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr) { char parentname[ZFS_MAX_DATASET_NAME_LEN]; int error; if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_RENAME, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); if ((error = zfs_get_parent(to, parentname, sizeof (parentname))) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (error); } static int zfs_secpolicy_rename(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; return (zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr)); } static int zfs_secpolicy_promote(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; dsl_pool_t *dp; dsl_dataset_t *clone; int error; error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_PROMOTE, cr); if (error != 0) return (error); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &clone); if (error == 0) { char parentname[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_t *origin = NULL; dsl_dir_t *dd; dd = clone->ds_dir; error = dsl_dataset_hold_obj(dd->dd_pool, dsl_dir_phys(dd)->dd_origin_obj, FTAG, &origin); if (error != 0) { dsl_dataset_rele(clone, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(zc->zc_name, clone, ZFS_DELEG_PERM_MOUNT, cr); dsl_dataset_name(origin, parentname); if (error == 0) { error = zfs_secpolicy_write_perms_ds(parentname, origin, ZFS_DELEG_PERM_PROMOTE, cr); } dsl_dataset_rele(clone, FTAG); dsl_dataset_rele(origin, FTAG); } dsl_pool_rele(dp, FTAG); return (error); } static int zfs_secpolicy_recv(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_RECEIVE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_CREATE, cr)); } int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr) { return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_SNAPSHOT, cr)); } /* * Check for permission to create each snapshot in the nvlist. */ static int zfs_secpolicy_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; nvlist_t *snaps; int error = 0; nvpair_t *pair; snaps = fnvlist_lookup_nvlist(innvl, "snaps"); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { char *name = nvpair_name(pair); char *atp = strchr(name, '@'); if (atp == NULL) { error = SET_ERROR(EINVAL); break; } *atp = '\0'; error = zfs_secpolicy_snapshot_perms(name, cr); *atp = '@'; if (error != 0) break; } return (error); } /* * Check for permission to create each bookmark in the nvlist. */ static int zfs_secpolicy_bookmark(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; int error = 0; for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_BOOKMARK, cr); *hashp = '#'; if (error != 0) break; } return (error); } static int zfs_secpolicy_destroy_bookmarks(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; nvpair_t *pair, *nextpair; int error = 0; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nextpair) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); nextpair = nvlist_next_nvpair(innvl, pair); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr); *hashp = '#'; if (error == ENOENT) { /* * Ignore any filesystems that don't exist (we consider * their bookmarks "already destroyed"). Remove * the name from the nvl here in case the filesystem * is created between now and when we try to destroy * the bookmark (in which case we don't want to * destroy it since we haven't checked for permission). */ fnvlist_remove_nvpair(innvl, pair); error = 0; } if (error != 0) break; } return (error); } static int zfs_secpolicy_log_history(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl, (void) cr; /* * Even root must have a proper TSD so that we know what pool * to log to. */ if (tsd_get(zfs_allow_log_key) == NULL) return (SET_ERROR(EPERM)); return (0); } static int zfs_secpolicy_create_clone(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char parentname[ZFS_MAX_DATASET_NAME_LEN]; int error; char *origin; if ((error = zfs_get_parent(zc->zc_name, parentname, sizeof (parentname))) != 0) return (error); if (nvlist_lookup_string(innvl, "origin", &origin) == 0 && (error = zfs_secpolicy_write_perms(origin, ZFS_DELEG_PERM_CLONE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)); } /* * Policy for pool operations - create/destroy pools, add vdevs, etc. Requires * SYS_CONFIG privilege, which is not available in a local zone. */ int zfs_secpolicy_config(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl; if (secpolicy_sys_config(cr, B_FALSE) != 0) return (SET_ERROR(EPERM)); return (0); } /* * Policy for object to name lookups. */ static int zfs_secpolicy_diff(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; int error; if ((error = secpolicy_sys_config(cr, B_FALSE)) == 0) return (0); error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr); return (error); } /* * Policy for fault injection. Requires all privileges. */ static int zfs_secpolicy_inject(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc, (void) innvl; return (secpolicy_zinject(cr)); } static int zfs_secpolicy_inherit_prop(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; zfs_prop_t prop = zfs_name_to_prop(zc->zc_value); if (prop == ZPROP_USERPROP) { if (!zfs_prop_user(zc->zc_value)) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_USERPROP, cr)); } else { return (zfs_secpolicy_setprop(zc->zc_name, prop, NULL, cr)); } } static int zfs_secpolicy_userspace_one(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); if (zc->zc_value[0] == 0) { /* * They are asking about a posix uid/gid. If it's * themself, allow it. */ if (zc->zc_objset_type == ZFS_PROP_USERUSED || zc->zc_objset_type == ZFS_PROP_USERQUOTA || zc->zc_objset_type == ZFS_PROP_USEROBJUSED || zc->zc_objset_type == ZFS_PROP_USEROBJQUOTA) { if (zc->zc_guid == crgetuid(cr)) return (0); } else if (zc->zc_objset_type == ZFS_PROP_GROUPUSED || zc->zc_objset_type == ZFS_PROP_GROUPQUOTA || zc->zc_objset_type == ZFS_PROP_GROUPOBJUSED || zc->zc_objset_type == ZFS_PROP_GROUPOBJQUOTA) { if (groupmember(zc->zc_guid, cr)) return (0); } /* else is for project quota/used */ } return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } static int zfs_secpolicy_userspace_many(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } static int zfs_secpolicy_userspace_upgrade(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) innvl; return (zfs_secpolicy_setprop(zc->zc_name, ZFS_PROP_VERSION, NULL, cr)); } static int zfs_secpolicy_hold(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; nvpair_t *pair; nvlist_t *holds; int error; holds = fnvlist_lookup_nvlist(innvl, "holds"); for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { char fsname[ZFS_MAX_DATASET_NAME_LEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_HOLD, cr); if (error != 0) return (error); } return (0); } static int zfs_secpolicy_release(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { (void) zc; nvpair_t *pair; int error; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char fsname[ZFS_MAX_DATASET_NAME_LEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_RELEASE, cr); if (error != 0) return (error); } return (0); } /* * Policy for allowing temporary snapshots to be taken or released */ static int zfs_secpolicy_tmp_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * A temporary snapshot is the same as a snapshot, * hold, destroy and release all rolled into one. * Delegated diff alone is sufficient that we allow this. */ int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr)) == 0) return (0); error = zfs_secpolicy_snapshot_perms(zc->zc_name, cr); if (innvl != NULL) { if (error == 0) error = zfs_secpolicy_hold(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_release(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_destroy(zc, innvl, cr); } return (error); } static int zfs_secpolicy_load_key(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_LOAD_KEY, cr)); } static int zfs_secpolicy_change_key(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_CHANGE_KEY, cr)); } /* * Returns the nvlist as specified by the user in the zfs_cmd_t. */ static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp) { char *packed; int error; nvlist_t *list = NULL; /* * Read in and unpack the user-supplied nvlist. */ if (size == 0) return (SET_ERROR(EINVAL)); packed = vmem_alloc(size, KM_SLEEP); if ((error = ddi_copyin((void *)(uintptr_t)nvl, packed, size, iflag)) != 0) { vmem_free(packed, size); return (SET_ERROR(EFAULT)); } if ((error = nvlist_unpack(packed, size, &list, 0)) != 0) { vmem_free(packed, size); return (error); } vmem_free(packed, size); *nvp = list; return (0); } /* * Reduce the size of this nvlist until it can be serialized in 'max' bytes. * Entries will be removed from the end of the nvlist, and one int32 entry * named "N_MORE_ERRORS" will be added indicating how many entries were * removed. */ static int nvlist_smush(nvlist_t *errors, size_t max) { size_t size; size = fnvlist_size(errors); if (size > max) { nvpair_t *more_errors; int n = 0; if (max < 1024) return (SET_ERROR(ENOMEM)); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, 0); more_errors = nvlist_prev_nvpair(errors, NULL); do { nvpair_t *pair = nvlist_prev_nvpair(errors, more_errors); fnvlist_remove_nvpair(errors, pair); n++; size = fnvlist_size(errors); } while (size > max); fnvlist_remove_nvpair(errors, more_errors); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, n); ASSERT3U(fnvlist_size(errors), <=, max); } return (0); } static int put_nvlist(zfs_cmd_t *zc, nvlist_t *nvl) { char *packed = NULL; int error = 0; size_t size; size = fnvlist_size(nvl); if (size > zc->zc_nvlist_dst_size) { error = SET_ERROR(ENOMEM); } else { packed = fnvlist_pack(nvl, &size); if (ddi_copyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst, size, zc->zc_iflags) != 0) error = SET_ERROR(EFAULT); fnvlist_pack_free(packed, size); } zc->zc_nvlist_dst_size = size; zc->zc_nvlist_dst_filled = B_TRUE; return (error); } int getzfsvfs_impl(objset_t *os, zfsvfs_t **zfvp) { int error = 0; if (dmu_objset_type(os) != DMU_OST_ZFS) { return (SET_ERROR(EINVAL)); } mutex_enter(&os->os_user_ptr_lock); *zfvp = dmu_objset_get_user(os); /* bump s_active only when non-zero to prevent umount race */ error = zfs_vfs_ref(zfvp); mutex_exit(&os->os_user_ptr_lock); return (error); } int getzfsvfs(const char *dsname, zfsvfs_t **zfvp) { objset_t *os; int error; error = dmu_objset_hold(dsname, FTAG, &os); if (error != 0) return (error); error = getzfsvfs_impl(os, zfvp); dmu_objset_rele(os, FTAG); return (error); } /* * Find a zfsvfs_t for a mounted filesystem, or create our own, in which * case its z_sb will be NULL, and it will be opened as the owner. * If 'writer' is set, the z_teardown_lock will be held for RW_WRITER, * which prevents all inode ops from running. */ static int zfsvfs_hold(const char *name, const void *tag, zfsvfs_t **zfvp, boolean_t writer) { int error = 0; if (getzfsvfs(name, zfvp) != 0) error = zfsvfs_create(name, B_FALSE, zfvp); if (error == 0) { if (writer) ZFS_TEARDOWN_ENTER_WRITE(*zfvp, tag); else ZFS_TEARDOWN_ENTER_READ(*zfvp, tag); if ((*zfvp)->z_unmounted) { /* * XXX we could probably try again, since the unmounting * thread should be just about to disassociate the * objset from the zfsvfs. */ ZFS_TEARDOWN_EXIT(*zfvp, tag); return (SET_ERROR(EBUSY)); } } return (error); } static void zfsvfs_rele(zfsvfs_t *zfsvfs, const void *tag) { ZFS_TEARDOWN_EXIT(zfsvfs, tag); if (zfs_vfs_held(zfsvfs)) { zfs_vfs_rele(zfsvfs); } else { dmu_objset_disown(zfsvfs->z_os, B_TRUE, zfsvfs); zfsvfs_free(zfsvfs); } } static int zfs_ioc_pool_create(zfs_cmd_t *zc) { int error; nvlist_t *config, *props = NULL; nvlist_t *rootprops = NULL; nvlist_t *zplprops = NULL; dsl_crypto_params_t *dcp = NULL; const char *spa_name = zc->zc_name; boolean_t unload_wkey = B_TRUE; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config))) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (props) { nvlist_t *nvl = NULL; nvlist_t *hidden_args = NULL; uint64_t version = SPA_VERSION; char *tname; (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version); if (!SPA_VERSION_IS_SUPPORTED(version)) { error = SET_ERROR(EINVAL); goto pool_props_bad; } (void) nvlist_lookup_nvlist(props, ZPOOL_ROOTFS_PROPS, &nvl); if (nvl) { error = nvlist_dup(nvl, &rootprops, KM_SLEEP); if (error != 0) goto pool_props_bad; (void) nvlist_remove_all(props, ZPOOL_ROOTFS_PROPS); } (void) nvlist_lookup_nvlist(props, ZPOOL_HIDDEN_ARGS, &hidden_args); error = dsl_crypto_params_create_nvlist(DCP_CMD_NONE, rootprops, hidden_args, &dcp); if (error != 0) goto pool_props_bad; (void) nvlist_remove_all(props, ZPOOL_HIDDEN_ARGS); VERIFY(nvlist_alloc(&zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops_root(version, rootprops, zplprops, NULL); if (error != 0) goto pool_props_bad; if (nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_TNAME), &tname) == 0) spa_name = tname; } error = spa_create(zc->zc_name, config, props, zplprops, dcp); /* * Set the remaining root properties */ if (!error && (error = zfs_set_prop_nvlist(spa_name, ZPROP_SRC_LOCAL, rootprops, NULL)) != 0) { (void) spa_destroy(spa_name); unload_wkey = B_FALSE; /* spa_destroy() unloads wrapping keys */ } pool_props_bad: nvlist_free(rootprops); nvlist_free(zplprops); nvlist_free(config); nvlist_free(props); dsl_crypto_params_free(dcp, unload_wkey && !!error); return (error); } static int zfs_ioc_pool_destroy(zfs_cmd_t *zc) { int error; zfs_log_history(zc); error = spa_destroy(zc->zc_name); return (error); } static int zfs_ioc_pool_import(zfs_cmd_t *zc) { nvlist_t *config, *props = NULL; uint64_t guid; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) != 0) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 || guid != zc->zc_guid) error = SET_ERROR(EINVAL); else error = spa_import(zc->zc_name, config, props, zc->zc_cookie); if (zc->zc_nvlist_dst != 0) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; } nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_pool_export(zfs_cmd_t *zc) { int error; boolean_t force = (boolean_t)zc->zc_cookie; boolean_t hardforce = (boolean_t)zc->zc_guid; zfs_log_history(zc); error = spa_export(zc->zc_name, NULL, force, hardforce); return (error); } static int zfs_ioc_pool_configs(zfs_cmd_t *zc) { nvlist_t *configs; int error; if ((configs = spa_all_configs(&zc->zc_cookie)) == NULL) return (SET_ERROR(EEXIST)); error = put_nvlist(zc, configs); nvlist_free(configs); return (error); } /* * inputs: * zc_name name of the pool * * outputs: * zc_cookie real errno * zc_nvlist_dst config nvlist * zc_nvlist_dst_size size of config nvlist */ static int zfs_ioc_pool_stats(zfs_cmd_t *zc) { nvlist_t *config; int error; int ret = 0; error = spa_get_stats(zc->zc_name, &config, zc->zc_value, sizeof (zc->zc_value)); if (config != NULL) { ret = put_nvlist(zc, config); nvlist_free(config); /* * The config may be present even if 'error' is non-zero. * In this case we return success, and preserve the real errno * in 'zc_cookie'. */ zc->zc_cookie = error; } else { ret = error; } return (ret); } /* * Try to import the given pool, returning pool stats as appropriate so that * user land knows which devices are available and overall pool health. */ static int zfs_ioc_pool_tryimport(zfs_cmd_t *zc) { nvlist_t *tryconfig, *config = NULL; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &tryconfig)) != 0) return (error); config = spa_tryimport(tryconfig); nvlist_free(tryconfig); if (config == NULL) return (SET_ERROR(EINVAL)); error = put_nvlist(zc, config); nvlist_free(config); return (error); } /* * inputs: * zc_name name of the pool * zc_cookie scan func (pool_scan_func_t) * zc_flags scrub pause/resume flag (pool_scrub_cmd_t) */ static int zfs_ioc_pool_scan(zfs_cmd_t *zc) { spa_t *spa; int error; if (zc->zc_flags >= POOL_SCRUB_FLAGS_END) return (SET_ERROR(EINVAL)); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_flags == POOL_SCRUB_PAUSE) error = spa_scrub_pause_resume(spa, POOL_SCRUB_PAUSE); else if (zc->zc_cookie == POOL_SCAN_NONE) error = spa_scan_stop(spa); else error = spa_scan(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_freeze(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { spa_freeze(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_pool_upgrade(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_cookie < spa_version(spa) || !SPA_VERSION_IS_SUPPORTED(zc->zc_cookie)) { spa_close(spa, FTAG); return (SET_ERROR(EINVAL)); } spa_upgrade(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_history(zfs_cmd_t *zc) { spa_t *spa; char *hist_buf; uint64_t size; int error; if ((size = zc->zc_history_len) == 0) return (SET_ERROR(EINVAL)); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } hist_buf = vmem_alloc(size, KM_SLEEP); if ((error = spa_history_get(spa, &zc->zc_history_offset, &zc->zc_history_len, hist_buf)) == 0) { error = ddi_copyout(hist_buf, (void *)(uintptr_t)zc->zc_history, zc->zc_history_len, zc->zc_iflags); } spa_close(spa, FTAG); vmem_free(hist_buf, size); return (error); } static int zfs_ioc_pool_reguid(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { error = spa_change_guid(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_dsobj_to_dsname(zfs_cmd_t *zc) { return (dsl_dsobj_to_dsname(zc->zc_name, zc->zc_obj, zc->zc_value)); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_value name of object */ static int zfs_ioc_obj_to_path(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele_flags(os, B_TRUE, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_path(os, zc->zc_obj, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele_flags(os, B_TRUE, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_stat stats on object * zc_value path to object */ static int zfs_ioc_obj_to_stats(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele_flags(os, B_TRUE, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_stats(os, zc->zc_obj, &zc->zc_stat, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele_flags(os, B_TRUE, FTAG); return (error); } static int zfs_ioc_vdev_add(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *config; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config); if (error == 0) { error = spa_vdev_add(spa, config); nvlist_free(config); } spa_close(spa, FTAG); return (error); } /* * inputs: * zc_name name of the pool * zc_guid guid of vdev to remove * zc_cookie cancel removal */ static int zfs_ioc_vdev_remove(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); if (zc->zc_cookie != 0) { error = spa_vdev_remove_cancel(spa); } else { error = spa_vdev_remove(spa, zc->zc_guid, B_FALSE); } spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_set_state(zfs_cmd_t *zc) { spa_t *spa; int error; vdev_state_t newstate = VDEV_STATE_UNKNOWN; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); switch (zc->zc_cookie) { case VDEV_STATE_ONLINE: error = vdev_online(spa, zc->zc_guid, zc->zc_obj, &newstate); break; case VDEV_STATE_OFFLINE: error = vdev_offline(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_FAULTED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL && zc->zc_obj != VDEV_AUX_EXTERNAL_PERSIST) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_fault(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_DEGRADED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_degrade(spa, zc->zc_guid, zc->zc_obj); break; + case VDEV_STATE_REMOVED: + error = vdev_remove_wanted(spa, zc->zc_guid); + break; + default: error = SET_ERROR(EINVAL); } zc->zc_cookie = newstate; spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_attach(zfs_cmd_t *zc) { spa_t *spa; nvlist_t *config; int replacing = zc->zc_cookie; int rebuild = zc->zc_simple; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) == 0) { error = spa_vdev_attach(spa, zc->zc_guid, config, replacing, rebuild); nvlist_free(config); } spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_detach(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_vdev_detach(spa, zc->zc_guid, 0, B_FALSE); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_split(zfs_cmd_t *zc) { spa_t *spa; nvlist_t *config, *props = NULL; int error; boolean_t exp = !!(zc->zc_cookie & ZPOOL_EXPORT_AFTER_SPLIT); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config))) { spa_close(spa, FTAG); return (error); } if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { spa_close(spa, FTAG); nvlist_free(config); return (error); } error = spa_vdev_split_mirror(spa, zc->zc_string, config, props, exp); spa_close(spa, FTAG); nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_vdev_setpath(zfs_cmd_t *zc) { spa_t *spa; const char *path = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setpath(spa, guid, path); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_setfru(zfs_cmd_t *zc) { spa_t *spa; const char *fru = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setfru(spa, guid, fru); spa_close(spa, FTAG); return (error); } static int zfs_ioc_objset_stats_impl(zfs_cmd_t *zc, objset_t *os) { int error = 0; nvlist_t *nv; dmu_objset_fast_stat(os, &zc->zc_objset_stats); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_all(os, &nv)) == 0) { dmu_objset_stats(os, nv); /* * NB: zvol_get_stats() will read the objset contents, * which we aren't supposed to do with a * DS_MODE_USER hold, because it could be * inconsistent. So this is a bit of a workaround... * XXX reading without owning */ if (!zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZVOL) { error = zvol_get_stats(os, nv); if (error == EIO) { nvlist_free(nv); return (error); } VERIFY0(error); } if (error == 0) error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_objset_stats(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error == 0) { error = zfs_ioc_objset_stats_impl(zc, os); dmu_objset_rele(os, FTAG); } return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_nvlist_dst received property nvlist * zc_nvlist_dst_size size of received property nvlist * * Gets received properties (distinct from local properties on or after * SPA_VERSION_RECVD_PROPS) for callers who want to differentiate received from * local property values. */ static int zfs_ioc_objset_recvd_props(zfs_cmd_t *zc) { int error = 0; nvlist_t *nv; /* * Without this check, we would return local property values if the * caller has not already received properties on or after * SPA_VERSION_RECVD_PROPS. */ if (!dsl_prop_get_hasrecvd(zc->zc_name)) return (SET_ERROR(ENOTSUP)); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_received(zc->zc_name, &nv)) == 0) { error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } static int nvl_add_zplprop(objset_t *os, nvlist_t *props, zfs_prop_t prop) { uint64_t value; int error; /* * zfs_get_zplprop() will either find a value or give us * the default value (if there is one). */ if ((error = zfs_get_zplprop(os, prop, &value)) != 0) return (error); VERIFY(nvlist_add_uint64(props, zfs_prop_to_name(prop), value) == 0); return (0); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for zpl property nvlist * * outputs: * zc_nvlist_dst zpl property nvlist * zc_nvlist_dst_size size of zpl property nvlist */ static int zfs_ioc_objset_zplprops(zfs_cmd_t *zc) { objset_t *os; int err; /* XXX reading without owning */ if ((err = dmu_objset_hold(zc->zc_name, FTAG, &os))) return (err); dmu_objset_fast_stat(os, &zc->zc_objset_stats); /* * NB: nvl_add_zplprop() will read the objset contents, * which we aren't supposed to do with a DS_MODE_USER * hold, because it could be inconsistent. */ if (zc->zc_nvlist_dst != 0 && !zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZFS) { nvlist_t *nv; VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0); if ((err = nvl_add_zplprop(os, nv, ZFS_PROP_VERSION)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_NORMALIZE)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_UTF8ONLY)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_CASE)) == 0) err = put_nvlist(zc, nv); nvlist_free(nv); } else { err = SET_ERROR(ENOENT); } dmu_objset_rele(os, FTAG); return (err); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_name name of next filesystem * zc_cookie zap cursor * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_dataset_list_next(zfs_cmd_t *zc) { objset_t *os; int error; char *p; size_t orig_len = strlen(zc->zc_name); top: if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os))) { if (error == ENOENT) error = SET_ERROR(ESRCH); return (error); } p = strrchr(zc->zc_name, '/'); if (p == NULL || p[1] != '\0') (void) strlcat(zc->zc_name, "/", sizeof (zc->zc_name)); p = zc->zc_name + strlen(zc->zc_name); do { error = dmu_dir_list_next(os, sizeof (zc->zc_name) - (p - zc->zc_name), p, NULL, &zc->zc_cookie); if (error == ENOENT) error = SET_ERROR(ESRCH); } while (error == 0 && zfs_dataset_name_hidden(zc->zc_name)); dmu_objset_rele(os, FTAG); /* * If it's an internal dataset (ie. with a '$' in its name), * don't try to get stats for it, otherwise we'll return ENOENT. */ if (error == 0 && strchr(zc->zc_name, '$') == NULL) { error = zfs_ioc_objset_stats(zc); /* fill in the stats */ if (error == ENOENT) { /* We lost a race with destroy, get the next one. */ zc->zc_name[orig_len] = '\0'; goto top; } } return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_src iteration range nvlist * zc_nvlist_src_size size of iteration range nvlist * * outputs: * zc_name name of next snapshot * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_snapshot_list_next(zfs_cmd_t *zc) { int error; objset_t *os, *ossnap; dsl_dataset_t *ds; uint64_t min_txg = 0, max_txg = 0; if (zc->zc_nvlist_src_size != 0) { nvlist_t *props = NULL; error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props); if (error != 0) return (error); (void) nvlist_lookup_uint64(props, SNAP_ITER_MIN_TXG, &min_txg); (void) nvlist_lookup_uint64(props, SNAP_ITER_MAX_TXG, &max_txg); nvlist_free(props); } error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) { return (error == ENOENT ? SET_ERROR(ESRCH) : error); } /* * A dataset name of maximum length cannot have any snapshots, * so exit immediately. */ if (strlcat(zc->zc_name, "@", sizeof (zc->zc_name)) >= ZFS_MAX_DATASET_NAME_LEN) { dmu_objset_rele(os, FTAG); return (SET_ERROR(ESRCH)); } while (error == 0) { if (issig(JUSTLOOKING) && issig(FORREAL)) { error = SET_ERROR(EINTR); break; } error = dmu_snapshot_list_next(os, sizeof (zc->zc_name) - strlen(zc->zc_name), zc->zc_name + strlen(zc->zc_name), &zc->zc_obj, &zc->zc_cookie, NULL); if (error == ENOENT) { error = SET_ERROR(ESRCH); break; } else if (error != 0) { break; } error = dsl_dataset_hold_obj(dmu_objset_pool(os), zc->zc_obj, FTAG, &ds); if (error != 0) break; if ((min_txg != 0 && dsl_get_creationtxg(ds) < min_txg) || (max_txg != 0 && dsl_get_creationtxg(ds) > max_txg)) { dsl_dataset_rele(ds, FTAG); /* undo snapshot name append */ *(strchr(zc->zc_name, '@') + 1) = '\0'; /* skip snapshot */ continue; } if (zc->zc_simple) { zc->zc_objset_stats.dds_creation_txg = dsl_get_creationtxg(ds); dsl_dataset_rele(ds, FTAG); break; } if ((error = dmu_objset_from_ds(ds, &ossnap)) != 0) { dsl_dataset_rele(ds, FTAG); break; } if ((error = zfs_ioc_objset_stats_impl(zc, ossnap)) != 0) { dsl_dataset_rele(ds, FTAG); break; } dsl_dataset_rele(ds, FTAG); break; } dmu_objset_rele(os, FTAG); /* if we failed, undo the @ that we tacked on to zc_name */ if (error != 0) *strchr(zc->zc_name, '@') = '\0'; return (error); } static int zfs_prop_set_userquota(const char *dsname, nvpair_t *pair) { const char *propname = nvpair_name(pair); uint64_t *valary; unsigned int vallen; const char *dash, *domain; zfs_userquota_prop_t type; uint64_t rid; uint64_t quota; zfsvfs_t *zfsvfs; int err; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) != 0) return (SET_ERROR(EINVAL)); } /* * A correctly constructed propname is encoded as * userquota@-. */ if ((dash = strchr(propname, '-')) == NULL || nvpair_value_uint64_array(pair, &valary, &vallen) != 0 || vallen != 3) return (SET_ERROR(EINVAL)); domain = dash + 1; type = valary[0]; rid = valary[1]; quota = valary[2]; err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_FALSE); if (err == 0) { err = zfs_set_userquota(zfsvfs, type, domain, rid, quota); zfsvfs_rele(zfsvfs, FTAG); } return (err); } /* * If the named property is one that has a special function to set its value, * return 0 on success and a positive error code on failure; otherwise if it is * not one of the special properties handled by this function, return -1. * * XXX: It would be better for callers of the property interface if we handled * these special cases in dsl_prop.c (in the dsl layer). */ static int zfs_prop_set_special(const char *dsname, zprop_source_t source, nvpair_t *pair) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval = 0; const char *strval = NULL; int err = -1; if (prop == ZPROP_USERPROP) { if (zfs_prop_userquota(propname)) return (zfs_prop_set_userquota(dsname, pair)); return (-1); } if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } /* all special properties are numeric except for keylocation */ if (zfs_prop_get_type(prop) == PROP_TYPE_STRING) { strval = fnvpair_value_string(pair); } else { intval = fnvpair_value_uint64(pair); } switch (prop) { case ZFS_PROP_QUOTA: err = dsl_dir_set_quota(dsname, source, intval); break; case ZFS_PROP_REFQUOTA: err = dsl_dataset_set_refquota(dsname, source, intval); break; case ZFS_PROP_FILESYSTEM_LIMIT: case ZFS_PROP_SNAPSHOT_LIMIT: if (intval == UINT64_MAX) { /* clearing the limit, just do it */ err = 0; } else { err = dsl_dir_activate_fs_ss_limit(dsname); } /* * Set err to -1 to force the zfs_set_prop_nvlist code down the * default path to set the value in the nvlist. */ if (err == 0) err = -1; break; case ZFS_PROP_KEYLOCATION: err = dsl_crypto_can_set_keylocation(dsname, strval); /* * Set err to -1 to force the zfs_set_prop_nvlist code down the * default path to set the value in the nvlist. */ if (err == 0) err = -1; break; case ZFS_PROP_RESERVATION: err = dsl_dir_set_reservation(dsname, source, intval); break; case ZFS_PROP_REFRESERVATION: err = dsl_dataset_set_refreservation(dsname, source, intval); break; case ZFS_PROP_COMPRESSION: err = dsl_dataset_set_compression(dsname, source, intval); /* * Set err to -1 to force the zfs_set_prop_nvlist code down the * default path to set the value in the nvlist. */ if (err == 0) err = -1; break; case ZFS_PROP_VOLSIZE: err = zvol_set_volsize(dsname, intval); break; case ZFS_PROP_SNAPDEV: err = zvol_set_snapdev(dsname, source, intval); break; case ZFS_PROP_VOLMODE: err = zvol_set_volmode(dsname, source, intval); break; case ZFS_PROP_VERSION: { zfsvfs_t *zfsvfs; if ((err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_TRUE)) != 0) break; err = zfs_set_version(zfsvfs, intval); zfsvfs_rele(zfsvfs, FTAG); if (err == 0 && intval >= ZPL_VERSION_USERSPACE) { zfs_cmd_t *zc; zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strlcpy(zc->zc_name, dsname, sizeof (zc->zc_name)); (void) zfs_ioc_userspace_upgrade(zc); (void) zfs_ioc_id_quota_upgrade(zc); kmem_free(zc, sizeof (zfs_cmd_t)); } break; } default: err = -1; } return (err); } static boolean_t zfs_is_namespace_prop(zfs_prop_t prop) { switch (prop) { case ZFS_PROP_ATIME: case ZFS_PROP_RELATIME: case ZFS_PROP_DEVICES: case ZFS_PROP_EXEC: case ZFS_PROP_SETUID: case ZFS_PROP_READONLY: case ZFS_PROP_XATTR: case ZFS_PROP_NBMAND: return (B_TRUE); default: return (B_FALSE); } } /* * This function is best effort. If it fails to set any of the given properties, * it continues to set as many as it can and returns the last error * encountered. If the caller provides a non-NULL errlist, it will be filled in * with the list of names of all the properties that failed along with the * corresponding error numbers. * * If every property is set successfully, zero is returned and errlist is not * modified. */ int zfs_set_prop_nvlist(const char *dsname, zprop_source_t source, nvlist_t *nvl, nvlist_t *errlist) { nvpair_t *pair; nvpair_t *propval; int rv = 0; int err; uint64_t intval; const char *strval; boolean_t should_update_mount_cache = B_FALSE; nvlist_t *genericnvl = fnvlist_alloc(); nvlist_t *retrynvl = fnvlist_alloc(); retry: pair = NULL; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); err = 0; /* decode the property value */ propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &propval) != 0) err = SET_ERROR(EINVAL); } /* Validate value type */ if (err == 0 && source == ZPROP_SRC_INHERITED) { /* inherited properties are expected to be booleans */ if (nvpair_type(propval) != DATA_TYPE_BOOLEAN) err = SET_ERROR(EINVAL); } else if (err == 0 && prop == ZPROP_USERPROP) { if (zfs_prop_user(propname)) { if (nvpair_type(propval) != DATA_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (zfs_prop_userquota(propname)) { if (nvpair_type(propval) != DATA_TYPE_UINT64_ARRAY) err = SET_ERROR(EINVAL); } else { err = SET_ERROR(EINVAL); } } else if (err == 0) { if (nvpair_type(propval) == DATA_TYPE_STRING) { if (zfs_prop_get_type(prop) != PROP_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (nvpair_type(propval) == DATA_TYPE_UINT64) { const char *unused; intval = fnvpair_value_uint64(propval); switch (zfs_prop_get_type(prop)) { case PROP_TYPE_NUMBER: break; case PROP_TYPE_STRING: err = SET_ERROR(EINVAL); break; case PROP_TYPE_INDEX: if (zfs_prop_index_to_string(prop, intval, &unused) != 0) err = SET_ERROR(ZFS_ERR_BADPROP); break; default: cmn_err(CE_PANIC, "unknown property type"); } } else { err = SET_ERROR(EINVAL); } } /* Validate permissions */ if (err == 0) err = zfs_check_settable(dsname, pair, CRED()); if (err == 0) { if (source == ZPROP_SRC_INHERITED) err = -1; /* does not need special handling */ else err = zfs_prop_set_special(dsname, source, pair); if (err == -1) { /* * For better performance we build up a list of * properties to set in a single transaction. */ err = nvlist_add_nvpair(genericnvl, pair); } else if (err != 0 && nvl != retrynvl) { /* * This may be a spurious error caused by * receiving quota and reservation out of order. * Try again in a second pass. */ err = nvlist_add_nvpair(retrynvl, pair); } } if (err != 0) { if (errlist != NULL) fnvlist_add_int32(errlist, propname, err); rv = err; } if (zfs_is_namespace_prop(prop)) should_update_mount_cache = B_TRUE; } if (nvl != retrynvl && !nvlist_empty(retrynvl)) { nvl = retrynvl; goto retry; } if (nvlist_empty(genericnvl)) goto out; /* * Try to set them all in one batch. */ err = dsl_props_set(dsname, source, genericnvl); if (err == 0) goto out; /* * If batching fails, we still want to set as many properties as we * can, so try setting them individually. */ pair = NULL; while ((pair = nvlist_next_nvpair(genericnvl, pair)) != NULL) { const char *propname = nvpair_name(pair); err = 0; propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); propval = fnvlist_lookup_nvpair(attrs, ZPROP_VALUE); } if (nvpair_type(propval) == DATA_TYPE_STRING) { strval = fnvpair_value_string(propval); err = dsl_prop_set_string(dsname, propname, source, strval); } else if (nvpair_type(propval) == DATA_TYPE_BOOLEAN) { err = dsl_prop_inherit(dsname, propname, source); } else { intval = fnvpair_value_uint64(propval); err = dsl_prop_set_int(dsname, propname, source, intval); } if (err != 0) { if (errlist != NULL) { fnvlist_add_int32(errlist, propname, err); } rv = err; } } out: if (should_update_mount_cache) zfs_ioctl_update_mount_cache(dsname); nvlist_free(genericnvl); nvlist_free(retrynvl); return (rv); } /* * Check that all the properties are valid user properties. */ static int zfs_check_userprops(nvlist_t *nvl) { nvpair_t *pair = NULL; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); if (!zfs_prop_user(propname) || nvpair_type(pair) != DATA_TYPE_STRING) return (SET_ERROR(EINVAL)); if (strlen(propname) >= ZAP_MAXNAMELEN) return (SET_ERROR(ENAMETOOLONG)); if (strlen(fnvpair_value_string(pair)) >= ZAP_MAXVALUELEN) return (SET_ERROR(E2BIG)); } return (0); } static void props_skip(nvlist_t *props, nvlist_t *skipped, nvlist_t **newprops) { nvpair_t *pair; VERIFY(nvlist_alloc(newprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); pair = NULL; while ((pair = nvlist_next_nvpair(props, pair)) != NULL) { if (nvlist_exists(skipped, nvpair_name(pair))) continue; VERIFY(nvlist_add_nvpair(*newprops, pair) == 0); } } static int clear_received_props(const char *dsname, nvlist_t *props, nvlist_t *skipped) { int err = 0; nvlist_t *cleared_props = NULL; props_skip(props, skipped, &cleared_props); if (!nvlist_empty(cleared_props)) { /* * Acts on local properties until the dataset has received * properties at least once on or after SPA_VERSION_RECVD_PROPS. */ zprop_source_t flags = (ZPROP_SRC_NONE | (dsl_prop_get_hasrecvd(dsname) ? ZPROP_SRC_RECEIVED : 0)); err = zfs_set_prop_nvlist(dsname, flags, cleared_props, NULL); } nvlist_free(cleared_props); return (err); } /* * inputs: * zc_name name of filesystem * zc_value name of property to set * zc_nvlist_src{_size} nvlist of properties to apply * zc_cookie received properties flag * * outputs: * zc_nvlist_dst{_size} error for each unapplied received property */ static int zfs_ioc_set_prop(zfs_cmd_t *zc) { nvlist_t *nvl; boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_RECEIVED : ZPROP_SRC_LOCAL); nvlist_t *errors; int error; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &nvl)) != 0) return (error); if (received) { nvlist_t *origprops; if (dsl_prop_get_received(zc->zc_name, &origprops) == 0) { (void) clear_received_props(zc->zc_name, origprops, nvl); nvlist_free(origprops); } error = dsl_prop_set_hasrecvd(zc->zc_name); } errors = fnvlist_alloc(); if (error == 0) error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, errors); if (zc->zc_nvlist_dst != 0 && errors != NULL) { (void) put_nvlist(zc, errors); } nvlist_free(errors); nvlist_free(nvl); return (error); } /* * inputs: * zc_name name of filesystem * zc_value name of property to inherit * zc_cookie revert to received value if TRUE * * outputs: none */ static int zfs_ioc_inherit_prop(zfs_cmd_t *zc) { const char *propname = zc->zc_value; zfs_prop_t prop = zfs_name_to_prop(propname); boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_NONE /* revert to received value, if any */ : ZPROP_SRC_INHERITED); /* explicitly inherit */ nvlist_t *dummy; nvpair_t *pair; zprop_type_t type; int err; if (!received) { /* * Only check this in the non-received case. We want to allow * 'inherit -S' to revert non-inheritable properties like quota * and reservation to the received or default values even though * they are not considered inheritable. */ if (prop != ZPROP_USERPROP && !zfs_prop_inheritable(prop)) return (SET_ERROR(EINVAL)); } if (prop == ZPROP_USERPROP) { if (!zfs_prop_user(propname)) return (SET_ERROR(EINVAL)); type = PROP_TYPE_STRING; } else if (prop == ZFS_PROP_VOLSIZE || prop == ZFS_PROP_VERSION) { return (SET_ERROR(EINVAL)); } else { type = zfs_prop_get_type(prop); } /* * zfs_prop_set_special() expects properties in the form of an * nvpair with type info. */ dummy = fnvlist_alloc(); switch (type) { case PROP_TYPE_STRING: VERIFY(0 == nvlist_add_string(dummy, propname, "")); break; case PROP_TYPE_NUMBER: case PROP_TYPE_INDEX: VERIFY(0 == nvlist_add_uint64(dummy, propname, 0)); break; default: err = SET_ERROR(EINVAL); goto errout; } pair = nvlist_next_nvpair(dummy, NULL); if (pair == NULL) { err = SET_ERROR(EINVAL); } else { err = zfs_prop_set_special(zc->zc_name, source, pair); if (err == -1) /* property is not "special", needs handling */ err = dsl_prop_inherit(zc->zc_name, zc->zc_value, source); } errout: nvlist_free(dummy); return (err); } static int zfs_ioc_pool_set_props(zfs_cmd_t *zc) { nvlist_t *props; spa_t *spa; int error; nvpair_t *pair; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) return (error); /* * If the only property is the configfile, then just do a spa_lookup() * to handle the faulted case. */ pair = nvlist_next_nvpair(props, NULL); if (pair != NULL && strcmp(nvpair_name(pair), zpool_prop_to_name(ZPOOL_PROP_CACHEFILE)) == 0 && nvlist_next_nvpair(props, pair) == NULL) { mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) { spa_configfile_set(spa, props, B_FALSE); - spa_write_cachefile(spa, B_FALSE, B_TRUE); + spa_write_cachefile(spa, B_FALSE, B_TRUE, B_FALSE); } mutex_exit(&spa_namespace_lock); if (spa != NULL) { nvlist_free(props); return (0); } } if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { nvlist_free(props); return (error); } error = spa_prop_set(spa, props); nvlist_free(props); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_props(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *nvp = NULL; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { /* * If the pool is faulted, there may be properties we can still * get (such as altroot and cachefile), so attempt to get them * anyway. */ mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) error = spa_prop_get(spa, &nvp); mutex_exit(&spa_namespace_lock); } else { error = spa_prop_get(spa, &nvp); spa_close(spa, FTAG); } if (error == 0 && zc->zc_nvlist_dst != 0) error = put_nvlist(zc, nvp); else error = SET_ERROR(EFAULT); nvlist_free(nvp); return (error); } /* * innvl: { * "vdevprops_set_vdev" -> guid * "vdevprops_set_props" -> { prop -> value } * } * * outnvl: propname -> error code (int32) */ static const zfs_ioc_key_t zfs_keys_vdev_set_props[] = { {ZPOOL_VDEV_PROPS_SET_VDEV, DATA_TYPE_UINT64, 0}, {ZPOOL_VDEV_PROPS_SET_PROPS, DATA_TYPE_NVLIST, 0} }; static int zfs_ioc_vdev_set_props(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { spa_t *spa; int error; vdev_t *vd; uint64_t vdev_guid; /* Early validation */ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_SET_VDEV, &vdev_guid) != 0) return (SET_ERROR(EINVAL)); if (outnvl == NULL) return (SET_ERROR(EINVAL)); if ((error = spa_open(poolname, &spa, FTAG)) != 0) return (error); ASSERT(spa_writeable(spa)); if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL) { spa_close(spa, FTAG); return (SET_ERROR(ENOENT)); } error = vdev_prop_set(vd, innvl, outnvl); spa_close(spa, FTAG); return (error); } /* * innvl: { * "vdevprops_get_vdev" -> guid * (optional) "vdevprops_get_props" -> { propname -> propid } * } * * outnvl: propname -> value */ static const zfs_ioc_key_t zfs_keys_vdev_get_props[] = { {ZPOOL_VDEV_PROPS_GET_VDEV, DATA_TYPE_UINT64, 0}, {ZPOOL_VDEV_PROPS_GET_PROPS, DATA_TYPE_NVLIST, ZK_OPTIONAL} }; static int zfs_ioc_vdev_get_props(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { spa_t *spa; int error; vdev_t *vd; uint64_t vdev_guid; /* Early validation */ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_GET_VDEV, &vdev_guid) != 0) return (SET_ERROR(EINVAL)); if (outnvl == NULL) return (SET_ERROR(EINVAL)); if ((error = spa_open(poolname, &spa, FTAG)) != 0) return (error); if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL) { spa_close(spa, FTAG); return (SET_ERROR(ENOENT)); } error = vdev_prop_get(vd, innvl, outnvl); spa_close(spa, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_src{_size} nvlist of delegated permissions * zc_perm_action allow/unallow flag * * outputs: none */ static int zfs_ioc_set_fsacl(zfs_cmd_t *zc) { int error; nvlist_t *fsaclnv = NULL; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &fsaclnv)) != 0) return (error); /* * Verify nvlist is constructed correctly */ if ((error = zfs_deleg_verify_nvlist(fsaclnv)) != 0) { nvlist_free(fsaclnv); return (SET_ERROR(EINVAL)); } /* * If we don't have PRIV_SYS_MOUNT, then validate * that user is allowed to hand out each permission in * the nvlist(s) */ error = secpolicy_zfs(CRED()); if (error != 0) { if (zc->zc_perm_action == B_FALSE) { error = dsl_deleg_can_allow(zc->zc_name, fsaclnv, CRED()); } else { error = dsl_deleg_can_unallow(zc->zc_name, fsaclnv, CRED()); } } if (error == 0) error = dsl_deleg_set(zc->zc_name, fsaclnv, zc->zc_perm_action); nvlist_free(fsaclnv); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * zc_nvlist_src{_size} nvlist of delegated permissions */ static int zfs_ioc_get_fsacl(zfs_cmd_t *zc) { nvlist_t *nvp; int error; if ((error = dsl_deleg_get(zc->zc_name, &nvp)) == 0) { error = put_nvlist(zc, nvp); nvlist_free(nvp); } return (error); } static void zfs_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { zfs_creat_t *zct = arg; zfs_create_fs(os, cr, zct->zct_zplprops, tx); } #define ZFS_PROP_UNDEFINED ((uint64_t)-1) /* * inputs: * os parent objset pointer (NULL if root fs) * fuids_ok fuids allowed in this version of the spa? * sa_ok SAs allowed in this version of the spa? * createprops list of properties requested by creator * * outputs: * zplprops values for the zplprops we attach to the master node object * is_ci true if requested file system will be purely case-insensitive * * Determine the settings for utf8only, normalization and * casesensitivity. Specific values may have been requested by the * creator and/or we can inherit values from the parent dataset. If * the file system is of too early a vintage, a creator can not * request settings for these properties, even if the requested * setting is the default value. We don't actually want to create dsl * properties for these, so remove them from the source nvlist after * processing. */ static int zfs_fill_zplprops_impl(objset_t *os, uint64_t zplver, boolean_t fuids_ok, boolean_t sa_ok, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { uint64_t sense = ZFS_PROP_UNDEFINED; uint64_t norm = ZFS_PROP_UNDEFINED; uint64_t u8 = ZFS_PROP_UNDEFINED; int error; ASSERT(zplprops != NULL); /* parent dataset must be a filesystem */ if (os != NULL && os->os_phys->os_type != DMU_OST_ZFS) return (SET_ERROR(ZFS_ERR_WRONG_PARENT)); /* * Pull out creator prop choices, if any. */ if (createprops) { (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_VERSION), &zplver); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), &norm); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), &u8); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_CASE), &sense); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_CASE)); } /* * If the zpl version requested is whacky or the file system * or pool is version is too "young" to support normalization * and the creator tried to set a value for one of the props, * error out. */ if ((zplver < ZPL_VERSION_INITIAL || zplver > ZPL_VERSION) || (zplver >= ZPL_VERSION_FUID && !fuids_ok) || (zplver >= ZPL_VERSION_SA && !sa_ok) || (zplver < ZPL_VERSION_NORMALIZATION && (norm != ZFS_PROP_UNDEFINED || u8 != ZFS_PROP_UNDEFINED || sense != ZFS_PROP_UNDEFINED))) return (SET_ERROR(ENOTSUP)); /* * Put the version in the zplprops */ VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_VERSION), zplver) == 0); if (norm == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &norm)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), norm) == 0); /* * If we're normalizing, names must always be valid UTF-8 strings. */ if (norm) u8 = 1; if (u8 == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &u8)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), u8) == 0); if (sense == ZFS_PROP_UNDEFINED && (error = zfs_get_zplprop(os, ZFS_PROP_CASE, &sense)) != 0) return (error); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_CASE), sense) == 0); if (is_ci) *is_ci = (sense == ZFS_CASE_INSENSITIVE); return (0); } static int zfs_fill_zplprops(const char *dataset, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok, sa_ok; uint64_t zplver = ZPL_VERSION; objset_t *os = NULL; char parentname[ZFS_MAX_DATASET_NAME_LEN]; spa_t *spa; uint64_t spa_vers; int error; zfs_get_parent(dataset, parentname, sizeof (parentname)); if ((error = spa_open(dataset, &spa, FTAG)) != 0) return (error); spa_vers = spa_version(spa); spa_close(spa, FTAG); zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); /* * Open parent object set so we can inherit zplprop values. */ if ((error = dmu_objset_hold(parentname, FTAG, &os)) != 0) return (error); error = zfs_fill_zplprops_impl(os, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); dmu_objset_rele(os, FTAG); return (error); } static int zfs_fill_zplprops_root(uint64_t spa_vers, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok; boolean_t sa_ok; uint64_t zplver = ZPL_VERSION; int error; zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); error = zfs_fill_zplprops_impl(NULL, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); return (error); } /* * innvl: { * "type" -> dmu_objset_type_t (int32) * (optional) "props" -> { prop -> value } * (optional) "hidden_args" -> { "wkeydata" -> value } * raw uint8_t array of encryption wrapping key data (32 bytes) * } * * outnvl: propname -> error code (int32) */ static const zfs_ioc_key_t zfs_keys_create[] = { {"type", DATA_TYPE_INT32, 0}, {"props", DATA_TYPE_NVLIST, ZK_OPTIONAL}, {"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL}, }; static int zfs_ioc_create(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; zfs_creat_t zct = { 0 }; nvlist_t *nvprops = NULL; nvlist_t *hidden_args = NULL; void (*cbfunc)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); dmu_objset_type_t type; boolean_t is_insensitive = B_FALSE; dsl_crypto_params_t *dcp = NULL; type = (dmu_objset_type_t)fnvlist_lookup_int32(innvl, "type"); (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); (void) nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args); switch (type) { case DMU_OST_ZFS: cbfunc = zfs_create_cb; break; case DMU_OST_ZVOL: cbfunc = zvol_create_cb; break; default: cbfunc = NULL; break; } if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); zct.zct_props = nvprops; if (cbfunc == NULL) return (SET_ERROR(EINVAL)); if (type == DMU_OST_ZVOL) { uint64_t volsize, volblocksize; if (nvprops == NULL) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) != 0) return (SET_ERROR(EINVAL)); if ((error = nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize)) != 0 && error != ENOENT) return (SET_ERROR(EINVAL)); if (error != 0) volblocksize = zfs_prop_default_numeric( ZFS_PROP_VOLBLOCKSIZE); if ((error = zvol_check_volblocksize(fsname, volblocksize)) != 0 || (error = zvol_check_volsize(volsize, volblocksize)) != 0) return (error); } else if (type == DMU_OST_ZFS) { int error; /* * We have to have normalization and * case-folding flags correct when we do the * file system creation, so go figure them out * now. */ VERIFY(nvlist_alloc(&zct.zct_zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops(fsname, nvprops, zct.zct_zplprops, &is_insensitive); if (error != 0) { nvlist_free(zct.zct_zplprops); return (error); } } error = dsl_crypto_params_create_nvlist(DCP_CMD_NONE, nvprops, hidden_args, &dcp); if (error != 0) { nvlist_free(zct.zct_zplprops); return (error); } error = dmu_objset_create(fsname, type, is_insensitive ? DS_FLAG_CI_DATASET : 0, dcp, cbfunc, &zct); nvlist_free(zct.zct_zplprops); dsl_crypto_params_free(dcp, !!error); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) { spa_t *spa; int error2; /* * Volumes will return EBUSY and cannot be destroyed * until all asynchronous minor handling (e.g. from * setting the volmode property) has completed. Wait for * the spa_zvol_taskq to drain then retry. */ error2 = dsl_destroy_head(fsname); while ((error2 == EBUSY) && (type == DMU_OST_ZVOL)) { error2 = spa_open(fsname, &spa, FTAG); if (error2 == 0) { taskq_wait(spa->spa_zvol_taskq); spa_close(spa, FTAG); } error2 = dsl_destroy_head(fsname); } } } return (error); } /* * innvl: { * "origin" -> name of origin snapshot * (optional) "props" -> { prop -> value } * (optional) "hidden_args" -> { "wkeydata" -> value } * raw uint8_t array of encryption wrapping key data (32 bytes) * } * * outputs: * outnvl: propname -> error code (int32) */ static const zfs_ioc_key_t zfs_keys_clone[] = { {"origin", DATA_TYPE_STRING, 0}, {"props", DATA_TYPE_NVLIST, ZK_OPTIONAL}, {"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL}, }; static int zfs_ioc_clone(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; nvlist_t *nvprops = NULL; const char *origin_name; origin_name = fnvlist_lookup_string(innvl, "origin"); (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); if (dataset_namecheck(origin_name, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); error = dmu_objset_clone(fsname, origin_name); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) (void) dsl_destroy_head(fsname); } return (error); } static const zfs_ioc_key_t zfs_keys_remap[] = { /* no nvl keys */ }; static int zfs_ioc_remap(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { /* This IOCTL is no longer supported. */ (void) fsname, (void) innvl, (void) outnvl; return (0); } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional) "props" -> { prop -> value (string) } * } * * outnvl: snapshot -> error code (int32) */ static const zfs_ioc_key_t zfs_keys_snapshot[] = { {"snaps", DATA_TYPE_NVLIST, 0}, {"props", DATA_TYPE_NVLIST, ZK_OPTIONAL}, }; static int zfs_ioc_snapshot(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { nvlist_t *snaps; nvlist_t *props = NULL; int error, poollen; nvpair_t *pair; (void) nvlist_lookup_nvlist(innvl, "props", &props); if (!nvlist_empty(props) && zfs_earlier_version(poolname, SPA_VERSION_SNAP_PROPS)) return (SET_ERROR(ENOTSUP)); if ((error = zfs_check_userprops(props)) != 0) return (error); snaps = fnvlist_lookup_nvlist(innvl, "snaps"); poollen = strlen(poolname); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); char *cp = strchr(name, '@'); /* * The snap name must contain an @, and the part after it must * contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The snap must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '@')) return (SET_ERROR(EXDEV)); /* * Check for permission to set the properties on the fs. */ if (!nvlist_empty(props)) { *cp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_USERPROP, CRED()); *cp = '@'; if (error != 0) return (error); } /* This must be the only snap of this fs. */ for (nvpair_t *pair2 = nvlist_next_nvpair(snaps, pair); pair2 != NULL; pair2 = nvlist_next_nvpair(snaps, pair2)) { if (strncmp(name, nvpair_name(pair2), cp - name + 1) == 0) { return (SET_ERROR(EXDEV)); } } } error = dsl_dataset_snapshot(snaps, props, outnvl); return (error); } /* * innvl: "message" -> string */ static const zfs_ioc_key_t zfs_keys_log_history[] = { {"message", DATA_TYPE_STRING, 0}, }; static int zfs_ioc_log_history(const char *unused, nvlist_t *innvl, nvlist_t *outnvl) { (void) unused, (void) outnvl; const char *message; char *poolname; spa_t *spa; int error; /* * The poolname in the ioctl is not set, we get it from the TSD, * which was set at the end of the last successful ioctl that allows * logging. The secpolicy func already checked that it is set. * Only one log ioctl is allowed after each successful ioctl, so * we clear the TSD here. */ poolname = tsd_get(zfs_allow_log_key); if (poolname == NULL) return (SET_ERROR(EINVAL)); (void) tsd_set(zfs_allow_log_key, NULL); error = spa_open(poolname, &spa, FTAG); kmem_strfree(poolname); if (error != 0) return (error); message = fnvlist_lookup_string(innvl, "message"); if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } error = spa_history_log(spa, message); spa_close(spa, FTAG); return (error); } /* * This ioctl is used to set the bootenv configuration on the current * pool. This configuration is stored in the second padding area of the label, * and it is used by the bootloader(s) to store the bootloader and/or system * specific data. * The data is stored as nvlist data stream, and is protected by * an embedded checksum. * The version can have two possible values: * VB_RAW: nvlist should have key GRUB_ENVMAP, value DATA_TYPE_STRING. * VB_NVLIST: nvlist with arbitrary pairs. */ static const zfs_ioc_key_t zfs_keys_set_bootenv[] = { {"version", DATA_TYPE_UINT64, 0}, {"", DATA_TYPE_ANY, ZK_OPTIONAL | ZK_WILDCARDLIST}, }; static int zfs_ioc_set_bootenv(const char *name, nvlist_t *innvl, nvlist_t *outnvl) { int error; spa_t *spa; if ((error = spa_open(name, &spa, FTAG)) != 0) return (error); spa_vdev_state_enter(spa, SCL_ALL); error = vdev_label_write_bootenv(spa->spa_root_vdev, innvl); (void) spa_vdev_state_exit(spa, NULL, 0); spa_close(spa, FTAG); return (error); } static const zfs_ioc_key_t zfs_keys_get_bootenv[] = { /* no nvl keys */ }; static int zfs_ioc_get_bootenv(const char *name, nvlist_t *innvl, nvlist_t *outnvl) { spa_t *spa; int error; if ((error = spa_open(name, &spa, FTAG)) != 0) return (error); spa_vdev_state_enter(spa, SCL_ALL); error = vdev_label_read_bootenv(spa->spa_root_vdev, outnvl); (void) spa_vdev_state_exit(spa, NULL, 0); spa_close(spa, FTAG); return (error); } /* * The dp_config_rwlock must not be held when calling this, because the * unmount may need to write out data. * * This function is best-effort. Callers must deal gracefully if it * remains mounted (or is remounted after this call). * * Returns 0 if the argument is not a snapshot, or it is not currently a * filesystem, or we were able to unmount it. Returns error code otherwise. */ void zfs_unmount_snap(const char *snapname) { if (strchr(snapname, '@') == NULL) return; (void) zfsctl_snapshot_unmount(snapname, MNT_FORCE); } static int zfs_unmount_snap_cb(const char *snapname, void *arg) { (void) arg; zfs_unmount_snap(snapname); return (0); } /* * When a clone is destroyed, its origin may also need to be destroyed, * in which case it must be unmounted. This routine will do that unmount * if necessary. */ void zfs_destroy_unmount_origin(const char *fsname) { int error; objset_t *os; dsl_dataset_t *ds; error = dmu_objset_hold(fsname, FTAG, &os); if (error != 0) return; ds = dmu_objset_ds(os); if (dsl_dir_is_clone(ds->ds_dir) && DS_IS_DEFER_DESTROY(ds->ds_prev)) { char originname[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_name(ds->ds_prev, originname); dmu_objset_rele(os, FTAG); zfs_unmount_snap(originname); } else { dmu_objset_rele(os, FTAG); } } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional boolean) "defer" * } * * outnvl: snapshot -> error code (int32) */ static const zfs_ioc_key_t zfs_keys_destroy_snaps[] = { {"snaps", DATA_TYPE_NVLIST, 0}, {"defer", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, }; static int zfs_ioc_destroy_snaps(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { int poollen; nvlist_t *snaps; nvpair_t *pair; boolean_t defer; spa_t *spa; snaps = fnvlist_lookup_nvlist(innvl, "snaps"); defer = nvlist_exists(innvl, "defer"); poollen = strlen(poolname); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); /* * The snap must be in the specified pool to prevent the * invalid removal of zvol minors below. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '@')) return (SET_ERROR(EXDEV)); zfs_unmount_snap(nvpair_name(pair)); if (spa_open(name, &spa, FTAG) == 0) { zvol_remove_minors(spa, name, B_TRUE); spa_close(spa, FTAG); } } return (dsl_destroy_snapshots_nvl(snaps, defer, outnvl)); } /* * Create bookmarks. The bookmark names are of the form #. * All bookmarks and snapshots must be in the same pool. * dsl_bookmark_create_nvl_validate describes the nvlist schema in more detail. * * innvl: { * new_bookmark1 -> existing_snapshot, * new_bookmark2 -> existing_bookmark, * } * * outnvl: bookmark -> error code (int32) * */ static const zfs_ioc_key_t zfs_keys_bookmark[] = { {"...", DATA_TYPE_STRING, ZK_WILDCARDLIST}, }; static int zfs_ioc_bookmark(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { (void) poolname; return (dsl_bookmark_create(innvl, outnvl)); } /* * innvl: { * property 1, property 2, ... * } * * outnvl: { * bookmark name 1 -> { property 1, property 2, ... }, * bookmark name 2 -> { property 1, property 2, ... } * } * */ static const zfs_ioc_key_t zfs_keys_get_bookmarks[] = { {"...", DATA_TYPE_BOOLEAN, ZK_WILDCARDLIST | ZK_OPTIONAL}, }; static int zfs_ioc_get_bookmarks(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { return (dsl_get_bookmarks(fsname, innvl, outnvl)); } /* * innvl is not used. * * outnvl: { * property 1, property 2, ... * } * */ static const zfs_ioc_key_t zfs_keys_get_bookmark_props[] = { /* no nvl keys */ }; static int zfs_ioc_get_bookmark_props(const char *bookmark, nvlist_t *innvl, nvlist_t *outnvl) { (void) innvl; char fsname[ZFS_MAX_DATASET_NAME_LEN]; char *bmname; bmname = strchr(bookmark, '#'); if (bmname == NULL) return (SET_ERROR(EINVAL)); bmname++; (void) strlcpy(fsname, bookmark, sizeof (fsname)); *(strchr(fsname, '#')) = '\0'; return (dsl_get_bookmark_props(fsname, bmname, outnvl)); } /* * innvl: { * bookmark name 1, bookmark name 2 * } * * outnvl: bookmark -> error code (int32) * */ static const zfs_ioc_key_t zfs_keys_destroy_bookmarks[] = { {"...", DATA_TYPE_BOOLEAN, ZK_WILDCARDLIST}, }; static int zfs_ioc_destroy_bookmarks(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { int error, poollen; poollen = strlen(poolname); for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { const char *name = nvpair_name(pair); const char *cp = strchr(name, '#'); /* * The bookmark name must contain an #, and the part after it * must contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The bookmark must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '#')) return (SET_ERROR(EXDEV)); } error = dsl_bookmark_destroy(innvl, outnvl); return (error); } static const zfs_ioc_key_t zfs_keys_channel_program[] = { {"program", DATA_TYPE_STRING, 0}, {"arg", DATA_TYPE_ANY, 0}, {"sync", DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL}, {"instrlimit", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"memlimit", DATA_TYPE_UINT64, ZK_OPTIONAL}, }; static int zfs_ioc_channel_program(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { char *program; uint64_t instrlimit, memlimit; boolean_t sync_flag; nvpair_t *nvarg = NULL; program = fnvlist_lookup_string(innvl, ZCP_ARG_PROGRAM); if (0 != nvlist_lookup_boolean_value(innvl, ZCP_ARG_SYNC, &sync_flag)) { sync_flag = B_TRUE; } if (0 != nvlist_lookup_uint64(innvl, ZCP_ARG_INSTRLIMIT, &instrlimit)) { instrlimit = ZCP_DEFAULT_INSTRLIMIT; } if (0 != nvlist_lookup_uint64(innvl, ZCP_ARG_MEMLIMIT, &memlimit)) { memlimit = ZCP_DEFAULT_MEMLIMIT; } nvarg = fnvlist_lookup_nvpair(innvl, ZCP_ARG_ARGLIST); if (instrlimit == 0 || instrlimit > zfs_lua_max_instrlimit) return (SET_ERROR(EINVAL)); if (memlimit == 0 || memlimit > zfs_lua_max_memlimit) return (SET_ERROR(EINVAL)); return (zcp_eval(poolname, program, sync_flag, instrlimit, memlimit, nvarg, outnvl)); } /* * innvl: unused * outnvl: empty */ static const zfs_ioc_key_t zfs_keys_pool_checkpoint[] = { /* no nvl keys */ }; static int zfs_ioc_pool_checkpoint(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { (void) innvl, (void) outnvl; return (spa_checkpoint(poolname)); } /* * innvl: unused * outnvl: empty */ static const zfs_ioc_key_t zfs_keys_pool_discard_checkpoint[] = { /* no nvl keys */ }; static int zfs_ioc_pool_discard_checkpoint(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { (void) innvl, (void) outnvl; return (spa_checkpoint_discard(poolname)); } /* * inputs: * zc_name name of dataset to destroy * zc_defer_destroy mark for deferred destroy * * outputs: none */ static int zfs_ioc_destroy(zfs_cmd_t *zc) { objset_t *os; dmu_objset_type_t ost; int err; err = dmu_objset_hold(zc->zc_name, FTAG, &os); if (err != 0) return (err); ost = dmu_objset_type(os); dmu_objset_rele(os, FTAG); if (ost == DMU_OST_ZFS) zfs_unmount_snap(zc->zc_name); if (strchr(zc->zc_name, '@')) { err = dsl_destroy_snapshot(zc->zc_name, zc->zc_defer_destroy); } else { err = dsl_destroy_head(zc->zc_name); if (err == EEXIST) { /* * It is possible that the given DS may have * hidden child (%recv) datasets - "leftovers" * resulting from the previously interrupted * 'zfs receive'. * * 6 extra bytes for /%recv */ char namebuf[ZFS_MAX_DATASET_NAME_LEN + 6]; if (snprintf(namebuf, sizeof (namebuf), "%s/%s", zc->zc_name, recv_clone_name) >= sizeof (namebuf)) return (SET_ERROR(EINVAL)); /* * Try to remove the hidden child (%recv) and after * that try to remove the target dataset. * If the hidden child (%recv) does not exist * the original error (EEXIST) will be returned */ err = dsl_destroy_head(namebuf); if (err == 0) err = dsl_destroy_head(zc->zc_name); else if (err == ENOENT) err = SET_ERROR(EEXIST); } } return (err); } /* * innvl: { * "initialize_command" -> POOL_INITIALIZE_{CANCEL|START|SUSPEND} (uint64) * "initialize_vdevs": { -> guids to initialize (nvlist) * "vdev_path_1": vdev_guid_1, (uint64), * "vdev_path_2": vdev_guid_2, (uint64), * ... * }, * } * * outnvl: { * "initialize_vdevs": { -> initialization errors (nvlist) * "vdev_path_1": errno, see function body for possible errnos (uint64) * "vdev_path_2": errno, ... (uint64) * ... * } * } * * EINVAL is returned for an unknown commands or if any of the provided vdev * guids have be specified with a type other than uint64. */ static const zfs_ioc_key_t zfs_keys_pool_initialize[] = { {ZPOOL_INITIALIZE_COMMAND, DATA_TYPE_UINT64, 0}, {ZPOOL_INITIALIZE_VDEVS, DATA_TYPE_NVLIST, 0} }; static int zfs_ioc_pool_initialize(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { uint64_t cmd_type; if (nvlist_lookup_uint64(innvl, ZPOOL_INITIALIZE_COMMAND, &cmd_type) != 0) { return (SET_ERROR(EINVAL)); } if (!(cmd_type == POOL_INITIALIZE_CANCEL || cmd_type == POOL_INITIALIZE_START || cmd_type == POOL_INITIALIZE_SUSPEND)) { return (SET_ERROR(EINVAL)); } nvlist_t *vdev_guids; if (nvlist_lookup_nvlist(innvl, ZPOOL_INITIALIZE_VDEVS, &vdev_guids) != 0) { return (SET_ERROR(EINVAL)); } for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL); pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) { uint64_t vdev_guid; if (nvpair_value_uint64(pair, &vdev_guid) != 0) { return (SET_ERROR(EINVAL)); } } spa_t *spa; int error = spa_open(poolname, &spa, FTAG); if (error != 0) return (error); nvlist_t *vdev_errlist = fnvlist_alloc(); int total_errors = spa_vdev_initialize(spa, vdev_guids, cmd_type, vdev_errlist); if (fnvlist_size(vdev_errlist) > 0) { fnvlist_add_nvlist(outnvl, ZPOOL_INITIALIZE_VDEVS, vdev_errlist); } fnvlist_free(vdev_errlist); spa_close(spa, FTAG); return (total_errors > 0 ? SET_ERROR(EINVAL) : 0); } /* * innvl: { * "trim_command" -> POOL_TRIM_{CANCEL|START|SUSPEND} (uint64) * "trim_vdevs": { -> guids to TRIM (nvlist) * "vdev_path_1": vdev_guid_1, (uint64), * "vdev_path_2": vdev_guid_2, (uint64), * ... * }, * "trim_rate" -> Target TRIM rate in bytes/sec. * "trim_secure" -> Set to request a secure TRIM. * } * * outnvl: { * "trim_vdevs": { -> TRIM errors (nvlist) * "vdev_path_1": errno, see function body for possible errnos (uint64) * "vdev_path_2": errno, ... (uint64) * ... * } * } * * EINVAL is returned for an unknown commands or if any of the provided vdev * guids have be specified with a type other than uint64. */ static const zfs_ioc_key_t zfs_keys_pool_trim[] = { {ZPOOL_TRIM_COMMAND, DATA_TYPE_UINT64, 0}, {ZPOOL_TRIM_VDEVS, DATA_TYPE_NVLIST, 0}, {ZPOOL_TRIM_RATE, DATA_TYPE_UINT64, ZK_OPTIONAL}, {ZPOOL_TRIM_SECURE, DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL}, }; static int zfs_ioc_pool_trim(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { uint64_t cmd_type; if (nvlist_lookup_uint64(innvl, ZPOOL_TRIM_COMMAND, &cmd_type) != 0) return (SET_ERROR(EINVAL)); if (!(cmd_type == POOL_TRIM_CANCEL || cmd_type == POOL_TRIM_START || cmd_type == POOL_TRIM_SUSPEND)) { return (SET_ERROR(EINVAL)); } nvlist_t *vdev_guids; if (nvlist_lookup_nvlist(innvl, ZPOOL_TRIM_VDEVS, &vdev_guids) != 0) return (SET_ERROR(EINVAL)); for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL); pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) { uint64_t vdev_guid; if (nvpair_value_uint64(pair, &vdev_guid) != 0) { return (SET_ERROR(EINVAL)); } } /* Optional, defaults to maximum rate when not provided */ uint64_t rate; if (nvlist_lookup_uint64(innvl, ZPOOL_TRIM_RATE, &rate) != 0) rate = 0; /* Optional, defaults to standard TRIM when not provided */ boolean_t secure; if (nvlist_lookup_boolean_value(innvl, ZPOOL_TRIM_SECURE, &secure) != 0) { secure = B_FALSE; } spa_t *spa; int error = spa_open(poolname, &spa, FTAG); if (error != 0) return (error); nvlist_t *vdev_errlist = fnvlist_alloc(); int total_errors = spa_vdev_trim(spa, vdev_guids, cmd_type, rate, !!zfs_trim_metaslab_skip, secure, vdev_errlist); if (fnvlist_size(vdev_errlist) > 0) fnvlist_add_nvlist(outnvl, ZPOOL_TRIM_VDEVS, vdev_errlist); fnvlist_free(vdev_errlist); spa_close(spa, FTAG); return (total_errors > 0 ? SET_ERROR(EINVAL) : 0); } /* * This ioctl waits for activity of a particular type to complete. If there is * no activity of that type in progress, it returns immediately, and the * returned value "waited" is false. If there is activity in progress, and no * tag is passed in, the ioctl blocks until all activity of that type is * complete, and then returns with "waited" set to true. * * If a tag is provided, it identifies a particular instance of an activity to * wait for. Currently, this is only valid for use with 'initialize', because * that is the only activity for which there can be multiple instances running * concurrently. In the case of 'initialize', the tag corresponds to the guid of * the vdev on which to wait. * * If a thread waiting in the ioctl receives a signal, the call will return * immediately, and the return value will be EINTR. * * innvl: { * "wait_activity" -> int32_t * (optional) "wait_tag" -> uint64_t * } * * outnvl: "waited" -> boolean_t */ static const zfs_ioc_key_t zfs_keys_pool_wait[] = { {ZPOOL_WAIT_ACTIVITY, DATA_TYPE_INT32, 0}, {ZPOOL_WAIT_TAG, DATA_TYPE_UINT64, ZK_OPTIONAL}, }; static int zfs_ioc_wait(const char *name, nvlist_t *innvl, nvlist_t *outnvl) { int32_t activity; uint64_t tag; boolean_t waited; int error; if (nvlist_lookup_int32(innvl, ZPOOL_WAIT_ACTIVITY, &activity) != 0) return (EINVAL); if (nvlist_lookup_uint64(innvl, ZPOOL_WAIT_TAG, &tag) == 0) error = spa_wait_tag(name, activity, tag, &waited); else error = spa_wait(name, activity, &waited); if (error == 0) fnvlist_add_boolean_value(outnvl, ZPOOL_WAIT_WAITED, waited); return (error); } /* * This ioctl waits for activity of a particular type to complete. If there is * no activity of that type in progress, it returns immediately, and the * returned value "waited" is false. If there is activity in progress, and no * tag is passed in, the ioctl blocks until all activity of that type is * complete, and then returns with "waited" set to true. * * If a thread waiting in the ioctl receives a signal, the call will return * immediately, and the return value will be EINTR. * * innvl: { * "wait_activity" -> int32_t * } * * outnvl: "waited" -> boolean_t */ static const zfs_ioc_key_t zfs_keys_fs_wait[] = { {ZFS_WAIT_ACTIVITY, DATA_TYPE_INT32, 0}, }; static int zfs_ioc_wait_fs(const char *name, nvlist_t *innvl, nvlist_t *outnvl) { int32_t activity; boolean_t waited = B_FALSE; int error; dsl_pool_t *dp; dsl_dir_t *dd; dsl_dataset_t *ds; if (nvlist_lookup_int32(innvl, ZFS_WAIT_ACTIVITY, &activity) != 0) return (SET_ERROR(EINVAL)); if (activity >= ZFS_WAIT_NUM_ACTIVITIES || activity < 0) return (SET_ERROR(EINVAL)); if ((error = dsl_pool_hold(name, FTAG, &dp)) != 0) return (error); if ((error = dsl_dataset_hold(dp, name, FTAG, &ds)) != 0) { dsl_pool_rele(dp, FTAG); return (error); } dd = ds->ds_dir; mutex_enter(&dd->dd_activity_lock); dd->dd_activity_waiters++; /* * We get a long-hold here so that the dsl_dataset_t and dsl_dir_t * aren't evicted while we're waiting. Normally this is prevented by * holding the pool, but we can't do that while we're waiting since * that would prevent TXGs from syncing out. Some of the functionality * of long-holds (e.g. preventing deletion) is unnecessary for this * case, since we would cancel the waiters before proceeding with a * deletion. An alternative mechanism for keeping the dataset around * could be developed but this is simpler. */ dsl_dataset_long_hold(ds, FTAG); dsl_pool_rele(dp, FTAG); error = dsl_dir_wait(dd, ds, activity, &waited); dsl_dataset_long_rele(ds, FTAG); dd->dd_activity_waiters--; if (dd->dd_activity_waiters == 0) cv_signal(&dd->dd_activity_cv); mutex_exit(&dd->dd_activity_lock); dsl_dataset_rele(ds, FTAG); if (error == 0) fnvlist_add_boolean_value(outnvl, ZFS_WAIT_WAITED, waited); return (error); } /* * fsname is name of dataset to rollback (to most recent snapshot) * * innvl may contain name of expected target snapshot * * outnvl: "target" -> name of most recent snapshot * } */ static const zfs_ioc_key_t zfs_keys_rollback[] = { {"target", DATA_TYPE_STRING, ZK_OPTIONAL}, }; static int zfs_ioc_rollback(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { zfsvfs_t *zfsvfs; zvol_state_handle_t *zv; char *target = NULL; int error; (void) nvlist_lookup_string(innvl, "target", &target); if (target != NULL) { const char *cp = strchr(target, '@'); /* * The snap name must contain an @, and the part after it must * contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); } if (getzfsvfs(fsname, &zfsvfs) == 0) { dsl_dataset_t *ds; ds = dmu_objset_ds(zfsvfs->z_os); error = zfs_suspend_fs(zfsvfs); if (error == 0) { int resume_err; error = dsl_dataset_rollback(fsname, target, zfsvfs, outnvl); resume_err = zfs_resume_fs(zfsvfs, ds); error = error ? error : resume_err; } zfs_vfs_rele(zfsvfs); } else if ((zv = zvol_suspend(fsname)) != NULL) { error = dsl_dataset_rollback(fsname, target, zvol_tag(zv), outnvl); zvol_resume(zv); } else { error = dsl_dataset_rollback(fsname, target, NULL, outnvl); } return (error); } static int recursive_unmount(const char *fsname, void *arg) { const char *snapname = arg; char *fullname; fullname = kmem_asprintf("%s@%s", fsname, snapname); zfs_unmount_snap(fullname); kmem_strfree(fullname); return (0); } /* * * snapname is the snapshot to redact. * innvl: { * "bookname" -> (string) * shortname of the redaction bookmark to generate * "snapnv" -> (nvlist, values ignored) * snapshots to redact snapname with respect to * } * * outnvl is unused */ static const zfs_ioc_key_t zfs_keys_redact[] = { {"bookname", DATA_TYPE_STRING, 0}, {"snapnv", DATA_TYPE_NVLIST, 0}, }; static int zfs_ioc_redact(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { (void) outnvl; nvlist_t *redactnvl = NULL; char *redactbook = NULL; if (nvlist_lookup_nvlist(innvl, "snapnv", &redactnvl) != 0) return (SET_ERROR(EINVAL)); if (fnvlist_num_pairs(redactnvl) == 0) return (SET_ERROR(ENXIO)); if (nvlist_lookup_string(innvl, "bookname", &redactbook) != 0) return (SET_ERROR(EINVAL)); return (dmu_redact_snap(snapname, redactnvl, redactbook)); } /* * inputs: * zc_name old name of dataset * zc_value new name of dataset * zc_cookie recursive flag (only valid for snapshots) * * outputs: none */ static int zfs_ioc_rename(zfs_cmd_t *zc) { objset_t *os; dmu_objset_type_t ost; boolean_t recursive = zc->zc_cookie & 1; boolean_t nounmount = !!(zc->zc_cookie & 2); char *at; int err; /* "zfs rename" from and to ...%recv datasets should both fail */ zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; zc->zc_value[sizeof (zc->zc_value) - 1] = '\0'; if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0 || dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_name, '%') || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); err = dmu_objset_hold(zc->zc_name, FTAG, &os); if (err != 0) return (err); ost = dmu_objset_type(os); dmu_objset_rele(os, FTAG); at = strchr(zc->zc_name, '@'); if (at != NULL) { /* snaps must be in same fs */ int error; if (strncmp(zc->zc_name, zc->zc_value, at - zc->zc_name + 1)) return (SET_ERROR(EXDEV)); *at = '\0'; if (ost == DMU_OST_ZFS && !nounmount) { error = dmu_objset_find(zc->zc_name, recursive_unmount, at + 1, recursive ? DS_FIND_CHILDREN : 0); if (error != 0) { *at = '@'; return (error); } } error = dsl_dataset_rename_snapshot(zc->zc_name, at + 1, strchr(zc->zc_value, '@') + 1, recursive); *at = '@'; return (error); } else { return (dsl_dir_rename(zc->zc_name, zc->zc_value)); } } static int zfs_check_settable(const char *dsname, nvpair_t *pair, cred_t *cr) { const char *propname = nvpair_name(pair); boolean_t issnap = (strchr(dsname, '@') != NULL); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval, compval; int err; if (prop == ZPROP_USERPROP) { if (zfs_prop_user(propname)) { if ((err = zfs_secpolicy_write_perms(dsname, ZFS_DELEG_PERM_USERPROP, cr))) return (err); return (0); } if (!issnap && zfs_prop_userquota(propname)) { const char *perm = NULL; const char *uq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA]; const char *gq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA]; const char *uiq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_USEROBJQUOTA]; const char *giq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_GROUPOBJQUOTA]; const char *pq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTQUOTA]; const char *piq_prefix = zfs_userquota_prop_prefixes[\ ZFS_PROP_PROJECTOBJQUOTA]; if (strncmp(propname, uq_prefix, strlen(uq_prefix)) == 0) { perm = ZFS_DELEG_PERM_USERQUOTA; } else if (strncmp(propname, uiq_prefix, strlen(uiq_prefix)) == 0) { perm = ZFS_DELEG_PERM_USEROBJQUOTA; } else if (strncmp(propname, gq_prefix, strlen(gq_prefix)) == 0) { perm = ZFS_DELEG_PERM_GROUPQUOTA; } else if (strncmp(propname, giq_prefix, strlen(giq_prefix)) == 0) { perm = ZFS_DELEG_PERM_GROUPOBJQUOTA; } else if (strncmp(propname, pq_prefix, strlen(pq_prefix)) == 0) { perm = ZFS_DELEG_PERM_PROJECTQUOTA; } else if (strncmp(propname, piq_prefix, strlen(piq_prefix)) == 0) { perm = ZFS_DELEG_PERM_PROJECTOBJQUOTA; } else { /* {USER|GROUP|PROJECT}USED are read-only */ return (SET_ERROR(EINVAL)); } if ((err = zfs_secpolicy_write_perms(dsname, perm, cr))) return (err); return (0); } return (SET_ERROR(EINVAL)); } if (issnap) return (SET_ERROR(EINVAL)); if (nvpair_type(pair) == DATA_TYPE_NVLIST) { /* * dsl_prop_get_all_impl() returns properties in this * format. */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } /* * Check that this value is valid for this pool version */ switch (prop) { case ZFS_PROP_COMPRESSION: /* * If the user specified gzip compression, make sure * the SPA supports it. We ignore any errors here since * we'll catch them later. */ if (nvpair_value_uint64(pair, &intval) == 0) { compval = ZIO_COMPRESS_ALGO(intval); if (compval >= ZIO_COMPRESS_GZIP_1 && compval <= ZIO_COMPRESS_GZIP_9 && zfs_earlier_version(dsname, SPA_VERSION_GZIP_COMPRESSION)) { return (SET_ERROR(ENOTSUP)); } if (compval == ZIO_COMPRESS_ZLE && zfs_earlier_version(dsname, SPA_VERSION_ZLE_COMPRESSION)) return (SET_ERROR(ENOTSUP)); if (compval == ZIO_COMPRESS_LZ4) { spa_t *spa; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LZ4_COMPRESS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } if (compval == ZIO_COMPRESS_ZSTD) { spa_t *spa; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_ZSTD_COMPRESS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } } break; case ZFS_PROP_COPIES: if (zfs_earlier_version(dsname, SPA_VERSION_DITTO_BLOCKS)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_VOLBLOCKSIZE: case ZFS_PROP_RECORDSIZE: /* Record sizes above 128k need the feature to be enabled */ if (nvpair_value_uint64(pair, &intval) == 0 && intval > SPA_OLD_MAXBLOCKSIZE) { spa_t *spa; /* * We don't allow setting the property above 1MB, * unless the tunable has been changed. */ if (intval > zfs_max_recordsize || intval > SPA_MAXBLOCKSIZE) return (SET_ERROR(ERANGE)); if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } break; case ZFS_PROP_DNODESIZE: /* Dnode sizes above 512 need the feature to be enabled */ if (nvpair_value_uint64(pair, &intval) == 0 && intval != ZFS_DNSIZE_LEGACY) { spa_t *spa; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } break; case ZFS_PROP_SPECIAL_SMALL_BLOCKS: /* * This property could require the allocation classes * feature to be active for setting, however we allow * it so that tests of settable properties succeed. * The CLI will issue a warning in this case. */ break; case ZFS_PROP_SHARESMB: if (zpl_earlier_version(dsname, ZPL_VERSION_FUID)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_ACLINHERIT: if (nvpair_type(pair) == DATA_TYPE_UINT64 && nvpair_value_uint64(pair, &intval) == 0) { if (intval == ZFS_ACL_PASSTHROUGH_X && zfs_earlier_version(dsname, SPA_VERSION_PASSTHROUGH_X)) return (SET_ERROR(ENOTSUP)); } break; case ZFS_PROP_CHECKSUM: case ZFS_PROP_DEDUP: { spa_feature_t feature; spa_t *spa; int err; /* dedup feature version checks */ if (prop == ZFS_PROP_DEDUP && zfs_earlier_version(dsname, SPA_VERSION_DEDUP)) return (SET_ERROR(ENOTSUP)); if (nvpair_type(pair) == DATA_TYPE_UINT64 && nvpair_value_uint64(pair, &intval) == 0) { /* check prop value is enabled in features */ feature = zio_checksum_to_feature( intval & ZIO_CHECKSUM_MASK); if (feature == SPA_FEATURE_NONE) break; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, feature)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } break; } default: break; } return (zfs_secpolicy_setprop(dsname, prop, pair, CRED())); } /* * Removes properties from the given props list that fail permission checks * needed to clear them and to restore them in case of a receive error. For each * property, make sure we have both set and inherit permissions. * * Returns the first error encountered if any permission checks fail. If the * caller provides a non-NULL errlist, it also gives the complete list of names * of all the properties that failed a permission check along with the * corresponding error numbers. The caller is responsible for freeing the * returned errlist. * * If every property checks out successfully, zero is returned and the list * pointed at by errlist is NULL. */ static int zfs_check_clearable(const char *dataset, nvlist_t *props, nvlist_t **errlist) { zfs_cmd_t *zc; nvpair_t *pair, *next_pair; nvlist_t *errors; int err, rv = 0; if (props == NULL) return (0); VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0); zc = kmem_alloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strlcpy(zc->zc_name, dataset, sizeof (zc->zc_name)); pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { next_pair = nvlist_next_nvpair(props, pair); (void) strlcpy(zc->zc_value, nvpair_name(pair), sizeof (zc->zc_value)); if ((err = zfs_check_settable(dataset, pair, CRED())) != 0 || (err = zfs_secpolicy_inherit_prop(zc, NULL, CRED())) != 0) { VERIFY(nvlist_remove_nvpair(props, pair) == 0); VERIFY(nvlist_add_int32(errors, zc->zc_value, err) == 0); } pair = next_pair; } kmem_free(zc, sizeof (zfs_cmd_t)); if ((pair = nvlist_next_nvpair(errors, NULL)) == NULL) { nvlist_free(errors); errors = NULL; } else { VERIFY(nvpair_value_int32(pair, &rv) == 0); } if (errlist == NULL) nvlist_free(errors); else *errlist = errors; return (rv); } static boolean_t propval_equals(nvpair_t *p1, nvpair_t *p2) { if (nvpair_type(p1) == DATA_TYPE_NVLIST) { /* dsl_prop_get_all_impl() format */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p1, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p1) == 0); } if (nvpair_type(p2) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p2, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p2) == 0); } if (nvpair_type(p1) != nvpair_type(p2)) return (B_FALSE); if (nvpair_type(p1) == DATA_TYPE_STRING) { char *valstr1, *valstr2; VERIFY(nvpair_value_string(p1, (char **)&valstr1) == 0); VERIFY(nvpair_value_string(p2, (char **)&valstr2) == 0); return (strcmp(valstr1, valstr2) == 0); } else { uint64_t intval1, intval2; VERIFY(nvpair_value_uint64(p1, &intval1) == 0); VERIFY(nvpair_value_uint64(p2, &intval2) == 0); return (intval1 == intval2); } } /* * Remove properties from props if they are not going to change (as determined * by comparison with origprops). Remove them from origprops as well, since we * do not need to clear or restore properties that won't change. */ static void props_reduce(nvlist_t *props, nvlist_t *origprops) { nvpair_t *pair, *next_pair; if (origprops == NULL) return; /* all props need to be received */ pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { const char *propname = nvpair_name(pair); nvpair_t *match; next_pair = nvlist_next_nvpair(props, pair); if ((nvlist_lookup_nvpair(origprops, propname, &match) != 0) || !propval_equals(pair, match)) goto next; /* need to set received value */ /* don't clear the existing received value */ (void) nvlist_remove_nvpair(origprops, match); /* don't bother receiving the property */ (void) nvlist_remove_nvpair(props, pair); next: pair = next_pair; } } /* * Extract properties that cannot be set PRIOR to the receipt of a dataset. * For example, refquota cannot be set until after the receipt of a dataset, * because in replication streams, an older/earlier snapshot may exceed the * refquota. We want to receive the older/earlier snapshot, but setting * refquota pre-receipt will set the dsl's ACTUAL quota, which will prevent * the older/earlier snapshot from being received (with EDQUOT). * * The ZFS test "zfs_receive_011_pos" demonstrates such a scenario. * * libzfs will need to be judicious handling errors encountered by props * extracted by this function. */ static nvlist_t * extract_delay_props(nvlist_t *props) { nvlist_t *delayprops; nvpair_t *nvp, *tmp; static const zfs_prop_t delayable[] = { ZFS_PROP_REFQUOTA, ZFS_PROP_KEYLOCATION, /* * Setting ZFS_PROP_SHARESMB requires the objset type to be * known, which is not possible prior to receipt of raw sends. */ ZFS_PROP_SHARESMB, 0 }; int i; VERIFY(nvlist_alloc(&delayprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (nvp = nvlist_next_nvpair(props, NULL); nvp != NULL; nvp = nvlist_next_nvpair(props, nvp)) { /* * strcmp() is safe because zfs_prop_to_name() always returns * a bounded string. */ for (i = 0; delayable[i] != 0; i++) { if (strcmp(zfs_prop_to_name(delayable[i]), nvpair_name(nvp)) == 0) { break; } } if (delayable[i] != 0) { tmp = nvlist_prev_nvpair(props, nvp); VERIFY(nvlist_add_nvpair(delayprops, nvp) == 0); VERIFY(nvlist_remove_nvpair(props, nvp) == 0); nvp = tmp; } } if (nvlist_empty(delayprops)) { nvlist_free(delayprops); delayprops = NULL; } return (delayprops); } static void zfs_allow_log_destroy(void *arg) { char *poolname = arg; if (poolname != NULL) kmem_strfree(poolname); } #ifdef ZFS_DEBUG static boolean_t zfs_ioc_recv_inject_err; #endif /* * nvlist 'errors' is always allocated. It will contain descriptions of * encountered errors, if any. It's the callers responsibility to free. */ static int zfs_ioc_recv_impl(char *tofs, char *tosnap, char *origin, nvlist_t *recvprops, nvlist_t *localprops, nvlist_t *hidden_args, boolean_t force, boolean_t heal, boolean_t resumable, int input_fd, dmu_replay_record_t *begin_record, uint64_t *read_bytes, uint64_t *errflags, nvlist_t **errors) { dmu_recv_cookie_t drc; int error = 0; int props_error = 0; offset_t off, noff; nvlist_t *local_delayprops = NULL; nvlist_t *recv_delayprops = NULL; nvlist_t *inherited_delayprops = NULL; nvlist_t *origprops = NULL; /* existing properties */ nvlist_t *origrecvd = NULL; /* existing received properties */ boolean_t first_recvd_props = B_FALSE; boolean_t tofs_was_redacted; zfs_file_t *input_fp; *read_bytes = 0; *errflags = 0; *errors = fnvlist_alloc(); off = 0; if ((input_fp = zfs_file_get(input_fd)) == NULL) return (SET_ERROR(EBADF)); noff = off = zfs_file_off(input_fp); error = dmu_recv_begin(tofs, tosnap, begin_record, force, heal, resumable, localprops, hidden_args, origin, &drc, input_fp, &off); if (error != 0) goto out; tofs_was_redacted = dsl_get_redacted(drc.drc_ds); /* * Set properties before we receive the stream so that they are applied * to the new data. Note that we must call dmu_recv_stream() if * dmu_recv_begin() succeeds. */ if (recvprops != NULL && !drc.drc_newfs) { if (spa_version(dsl_dataset_get_spa(drc.drc_ds)) >= SPA_VERSION_RECVD_PROPS && !dsl_prop_get_hasrecvd(tofs)) first_recvd_props = B_TRUE; /* * If new received properties are supplied, they are to * completely replace the existing received properties, * so stash away the existing ones. */ if (dsl_prop_get_received(tofs, &origrecvd) == 0) { nvlist_t *errlist = NULL; /* * Don't bother writing a property if its value won't * change (and avoid the unnecessary security checks). * * The first receive after SPA_VERSION_RECVD_PROPS is a * special case where we blow away all local properties * regardless. */ if (!first_recvd_props) props_reduce(recvprops, origrecvd); if (zfs_check_clearable(tofs, origrecvd, &errlist) != 0) (void) nvlist_merge(*errors, errlist, 0); nvlist_free(errlist); if (clear_received_props(tofs, origrecvd, first_recvd_props ? NULL : recvprops) != 0) *errflags |= ZPROP_ERR_NOCLEAR; } else { *errflags |= ZPROP_ERR_NOCLEAR; } } /* * Stash away existing properties so we can restore them on error unless * we're doing the first receive after SPA_VERSION_RECVD_PROPS, in which * case "origrecvd" will take care of that. */ if (localprops != NULL && !drc.drc_newfs && !first_recvd_props) { objset_t *os; if (dmu_objset_hold(tofs, FTAG, &os) == 0) { if (dsl_prop_get_all(os, &origprops) != 0) { *errflags |= ZPROP_ERR_NOCLEAR; } dmu_objset_rele(os, FTAG); } else { *errflags |= ZPROP_ERR_NOCLEAR; } } if (recvprops != NULL) { props_error = dsl_prop_set_hasrecvd(tofs); if (props_error == 0) { recv_delayprops = extract_delay_props(recvprops); (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, recvprops, *errors); } } if (localprops != NULL) { nvlist_t *oprops = fnvlist_alloc(); nvlist_t *xprops = fnvlist_alloc(); nvpair_t *nvp = NULL; while ((nvp = nvlist_next_nvpair(localprops, nvp)) != NULL) { if (nvpair_type(nvp) == DATA_TYPE_BOOLEAN) { /* -x property */ const char *name = nvpair_name(nvp); zfs_prop_t prop = zfs_name_to_prop(name); if (prop != ZPROP_USERPROP) { if (!zfs_prop_inheritable(prop)) continue; } else if (!zfs_prop_user(name)) continue; fnvlist_add_boolean(xprops, name); } else { /* -o property=value */ fnvlist_add_nvpair(oprops, nvp); } } local_delayprops = extract_delay_props(oprops); (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL, oprops, *errors); inherited_delayprops = extract_delay_props(xprops); (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_INHERITED, xprops, *errors); nvlist_free(oprops); nvlist_free(xprops); } error = dmu_recv_stream(&drc, &off); if (error == 0) { zfsvfs_t *zfsvfs = NULL; zvol_state_handle_t *zv = NULL; if (getzfsvfs(tofs, &zfsvfs) == 0) { /* online recv */ dsl_dataset_t *ds; int end_err; boolean_t stream_is_redacted = DMU_GET_FEATUREFLAGS( begin_record->drr_u.drr_begin. drr_versioninfo) & DMU_BACKUP_FEATURE_REDACTED; ds = dmu_objset_ds(zfsvfs->z_os); error = zfs_suspend_fs(zfsvfs); /* * If the suspend fails, then the recv_end will * likely also fail, and clean up after itself. */ end_err = dmu_recv_end(&drc, zfsvfs); /* * If the dataset was not redacted, but we received a * redacted stream onto it, we need to unmount the * dataset. Otherwise, resume the filesystem. */ if (error == 0 && !drc.drc_newfs && stream_is_redacted && !tofs_was_redacted) { error = zfs_end_fs(zfsvfs, ds); } else if (error == 0) { error = zfs_resume_fs(zfsvfs, ds); } error = error ? error : end_err; zfs_vfs_rele(zfsvfs); } else if ((zv = zvol_suspend(tofs)) != NULL) { error = dmu_recv_end(&drc, zvol_tag(zv)); zvol_resume(zv); } else { error = dmu_recv_end(&drc, NULL); } /* Set delayed properties now, after we're done receiving. */ if (recv_delayprops != NULL && error == 0) { (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, recv_delayprops, *errors); } if (local_delayprops != NULL && error == 0) { (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL, local_delayprops, *errors); } if (inherited_delayprops != NULL && error == 0) { (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_INHERITED, inherited_delayprops, *errors); } } /* * Merge delayed props back in with initial props, in case * we're DEBUG and zfs_ioc_recv_inject_err is set (which means * we have to make sure clear_received_props() includes * the delayed properties). * * Since zfs_ioc_recv_inject_err is only in DEBUG kernels, * using ASSERT() will be just like a VERIFY. */ if (recv_delayprops != NULL) { ASSERT(nvlist_merge(recvprops, recv_delayprops, 0) == 0); nvlist_free(recv_delayprops); } if (local_delayprops != NULL) { ASSERT(nvlist_merge(localprops, local_delayprops, 0) == 0); nvlist_free(local_delayprops); } if (inherited_delayprops != NULL) { ASSERT(nvlist_merge(localprops, inherited_delayprops, 0) == 0); nvlist_free(inherited_delayprops); } *read_bytes = off - noff; #ifdef ZFS_DEBUG if (zfs_ioc_recv_inject_err) { zfs_ioc_recv_inject_err = B_FALSE; error = 1; } #endif /* * On error, restore the original props. */ if (error != 0 && recvprops != NULL && !drc.drc_newfs) { if (clear_received_props(tofs, recvprops, NULL) != 0) { /* * We failed to clear the received properties. * Since we may have left a $recvd value on the * system, we can't clear the $hasrecvd flag. */ *errflags |= ZPROP_ERR_NORESTORE; } else if (first_recvd_props) { dsl_prop_unset_hasrecvd(tofs); } if (origrecvd == NULL && !drc.drc_newfs) { /* We failed to stash the original properties. */ *errflags |= ZPROP_ERR_NORESTORE; } /* * dsl_props_set() will not convert RECEIVED to LOCAL on or * after SPA_VERSION_RECVD_PROPS, so we need to specify LOCAL * explicitly if we're restoring local properties cleared in the * first new-style receive. */ if (origrecvd != NULL && zfs_set_prop_nvlist(tofs, (first_recvd_props ? ZPROP_SRC_LOCAL : ZPROP_SRC_RECEIVED), origrecvd, NULL) != 0) { /* * We stashed the original properties but failed to * restore them. */ *errflags |= ZPROP_ERR_NORESTORE; } } if (error != 0 && localprops != NULL && !drc.drc_newfs && !first_recvd_props) { nvlist_t *setprops; nvlist_t *inheritprops; nvpair_t *nvp; if (origprops == NULL) { /* We failed to stash the original properties. */ *errflags |= ZPROP_ERR_NORESTORE; goto out; } /* Restore original props */ setprops = fnvlist_alloc(); inheritprops = fnvlist_alloc(); nvp = NULL; while ((nvp = nvlist_next_nvpair(localprops, nvp)) != NULL) { const char *name = nvpair_name(nvp); const char *source; nvlist_t *attrs; if (!nvlist_exists(origprops, name)) { /* * Property was not present or was explicitly * inherited before the receive, restore this. */ fnvlist_add_boolean(inheritprops, name); continue; } attrs = fnvlist_lookup_nvlist(origprops, name); source = fnvlist_lookup_string(attrs, ZPROP_SOURCE); /* Skip received properties */ if (strcmp(source, ZPROP_SOURCE_VAL_RECVD) == 0) continue; if (strcmp(source, tofs) == 0) { /* Property was locally set */ fnvlist_add_nvlist(setprops, name, attrs); } else { /* Property was implicitly inherited */ fnvlist_add_boolean(inheritprops, name); } } if (zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL, setprops, NULL) != 0) *errflags |= ZPROP_ERR_NORESTORE; if (zfs_set_prop_nvlist(tofs, ZPROP_SRC_INHERITED, inheritprops, NULL) != 0) *errflags |= ZPROP_ERR_NORESTORE; nvlist_free(setprops); nvlist_free(inheritprops); } out: zfs_file_put(input_fp); nvlist_free(origrecvd); nvlist_free(origprops); if (error == 0) error = props_error; return (error); } /* * inputs: * zc_name name of containing filesystem (unused) * zc_nvlist_src{_size} nvlist of properties to apply * zc_nvlist_conf{_size} nvlist of properties to exclude * (DATA_TYPE_BOOLEAN) and override (everything else) * zc_value name of snapshot to create * zc_string name of clone origin (if DRR_FLAG_CLONE) * zc_cookie file descriptor to recv from * zc_begin_record the BEGIN record of the stream (not byteswapped) * zc_guid force flag * * outputs: * zc_cookie number of bytes read * zc_obj zprop_errflags_t * zc_nvlist_dst{_size} error for each unapplied received property */ static int zfs_ioc_recv(zfs_cmd_t *zc) { dmu_replay_record_t begin_record; nvlist_t *errors = NULL; nvlist_t *recvdprops = NULL; nvlist_t *localprops = NULL; char *origin = NULL; char *tosnap; char tofs[ZFS_MAX_DATASET_NAME_LEN]; int error = 0; if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_value, '@') == NULL || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); (void) strlcpy(tofs, zc->zc_value, sizeof (tofs)); tosnap = strchr(tofs, '@'); *tosnap++ = '\0'; if (zc->zc_nvlist_src != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &recvdprops)) != 0) return (error); if (zc->zc_nvlist_conf != 0 && (error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &localprops)) != 0) return (error); if (zc->zc_string[0]) origin = zc->zc_string; begin_record.drr_type = DRR_BEGIN; begin_record.drr_payloadlen = 0; begin_record.drr_u.drr_begin = zc->zc_begin_record; error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvdprops, localprops, NULL, zc->zc_guid, B_FALSE, B_FALSE, zc->zc_cookie, &begin_record, &zc->zc_cookie, &zc->zc_obj, &errors); nvlist_free(recvdprops); nvlist_free(localprops); /* * Now that all props, initial and delayed, are set, report the prop * errors to the caller. */ if (zc->zc_nvlist_dst_size != 0 && errors != NULL && (nvlist_smush(errors, zc->zc_nvlist_dst_size) != 0 || put_nvlist(zc, errors) != 0)) { /* * Caller made zc->zc_nvlist_dst less than the minimum expected * size or supplied an invalid address. */ error = SET_ERROR(EINVAL); } nvlist_free(errors); return (error); } /* * innvl: { * "snapname" -> full name of the snapshot to create * (optional) "props" -> received properties to set (nvlist) * (optional) "localprops" -> override and exclude properties (nvlist) * (optional) "origin" -> name of clone origin (DRR_FLAG_CLONE) * "begin_record" -> non-byteswapped dmu_replay_record_t * "input_fd" -> file descriptor to read stream from (int32) * (optional) "force" -> force flag (value ignored) * (optional) "heal" -> use send stream to heal data corruption * (optional) "resumable" -> resumable flag (value ignored) * (optional) "cleanup_fd" -> unused * (optional) "action_handle" -> unused * (optional) "hidden_args" -> { "wkeydata" -> value } * } * * outnvl: { * "read_bytes" -> number of bytes read * "error_flags" -> zprop_errflags_t * "errors" -> error for each unapplied received property (nvlist) * } */ static const zfs_ioc_key_t zfs_keys_recv_new[] = { {"snapname", DATA_TYPE_STRING, 0}, {"props", DATA_TYPE_NVLIST, ZK_OPTIONAL}, {"localprops", DATA_TYPE_NVLIST, ZK_OPTIONAL}, {"origin", DATA_TYPE_STRING, ZK_OPTIONAL}, {"begin_record", DATA_TYPE_BYTE_ARRAY, 0}, {"input_fd", DATA_TYPE_INT32, 0}, {"force", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"heal", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"resumable", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"cleanup_fd", DATA_TYPE_INT32, ZK_OPTIONAL}, {"action_handle", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL}, }; static int zfs_ioc_recv_new(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { dmu_replay_record_t *begin_record; uint_t begin_record_size; nvlist_t *errors = NULL; nvlist_t *recvprops = NULL; nvlist_t *localprops = NULL; nvlist_t *hidden_args = NULL; char *snapname; char *origin = NULL; char *tosnap; char tofs[ZFS_MAX_DATASET_NAME_LEN]; boolean_t force; boolean_t heal; boolean_t resumable; uint64_t read_bytes = 0; uint64_t errflags = 0; int input_fd = -1; int error; snapname = fnvlist_lookup_string(innvl, "snapname"); if (dataset_namecheck(snapname, NULL, NULL) != 0 || strchr(snapname, '@') == NULL || strchr(snapname, '%')) return (SET_ERROR(EINVAL)); (void) strlcpy(tofs, snapname, sizeof (tofs)); tosnap = strchr(tofs, '@'); *tosnap++ = '\0'; error = nvlist_lookup_string(innvl, "origin", &origin); if (error && error != ENOENT) return (error); error = nvlist_lookup_byte_array(innvl, "begin_record", (uchar_t **)&begin_record, &begin_record_size); if (error != 0 || begin_record_size != sizeof (*begin_record)) return (SET_ERROR(EINVAL)); input_fd = fnvlist_lookup_int32(innvl, "input_fd"); force = nvlist_exists(innvl, "force"); heal = nvlist_exists(innvl, "heal"); resumable = nvlist_exists(innvl, "resumable"); /* we still use "props" here for backwards compatibility */ error = nvlist_lookup_nvlist(innvl, "props", &recvprops); if (error && error != ENOENT) return (error); error = nvlist_lookup_nvlist(innvl, "localprops", &localprops); if (error && error != ENOENT) return (error); error = nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args); if (error && error != ENOENT) return (error); error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvprops, localprops, hidden_args, force, heal, resumable, input_fd, begin_record, &read_bytes, &errflags, &errors); fnvlist_add_uint64(outnvl, "read_bytes", read_bytes); fnvlist_add_uint64(outnvl, "error_flags", errflags); fnvlist_add_nvlist(outnvl, "errors", errors); nvlist_free(errors); nvlist_free(recvprops); nvlist_free(localprops); return (error); } typedef struct dump_bytes_io { zfs_file_t *dbi_fp; caddr_t dbi_buf; int dbi_len; int dbi_err; } dump_bytes_io_t; static void dump_bytes_cb(void *arg) { dump_bytes_io_t *dbi = (dump_bytes_io_t *)arg; zfs_file_t *fp; caddr_t buf; fp = dbi->dbi_fp; buf = dbi->dbi_buf; dbi->dbi_err = zfs_file_write(fp, buf, dbi->dbi_len, NULL); } static int dump_bytes(objset_t *os, void *buf, int len, void *arg) { dump_bytes_io_t dbi; dbi.dbi_fp = arg; dbi.dbi_buf = buf; dbi.dbi_len = len; #if defined(HAVE_LARGE_STACKS) dump_bytes_cb(&dbi); #else /* * The vn_rdwr() call is performed in a taskq to ensure that there is * always enough stack space to write safely to the target filesystem. * The ZIO_TYPE_FREE threads are used because there can be a lot of * them and they are used in vdev_file.c for a similar purpose. */ spa_taskq_dispatch_sync(dmu_objset_spa(os), ZIO_TYPE_FREE, ZIO_TASKQ_ISSUE, dump_bytes_cb, &dbi, TQ_SLEEP); #endif /* HAVE_LARGE_STACKS */ return (dbi.dbi_err); } /* * inputs: * zc_name name of snapshot to send * zc_cookie file descriptor to send stream to * zc_obj fromorigin flag (mutually exclusive with zc_fromobj) * zc_sendobj objsetid of snapshot to send * zc_fromobj objsetid of incremental fromsnap (may be zero) * zc_guid if set, estimate size of stream only. zc_cookie is ignored. * output size in zc_objset_type. * zc_flags lzc_send_flags * * outputs: * zc_objset_type estimated size, if zc_guid is set * * NOTE: This is no longer the preferred interface, any new functionality * should be added to zfs_ioc_send_new() instead. */ static int zfs_ioc_send(zfs_cmd_t *zc) { int error; offset_t off; boolean_t estimate = (zc->zc_guid != 0); boolean_t embedok = (zc->zc_flags & 0x1); boolean_t large_block_ok = (zc->zc_flags & 0x2); boolean_t compressok = (zc->zc_flags & 0x4); boolean_t rawok = (zc->zc_flags & 0x8); boolean_t savedok = (zc->zc_flags & 0x10); if (zc->zc_obj != 0) { dsl_pool_t *dp; dsl_dataset_t *tosnap; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (dsl_dir_is_clone(tosnap->ds_dir)) zc->zc_fromobj = dsl_dir_phys(tosnap->ds_dir)->dd_origin_obj; dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } if (estimate) { dsl_pool_t *dp; dsl_dataset_t *tosnap; dsl_dataset_t *fromsnap = NULL; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (zc->zc_fromobj != 0) { error = dsl_dataset_hold_obj(dp, zc->zc_fromobj, FTAG, &fromsnap); if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } } error = dmu_send_estimate_fast(tosnap, fromsnap, NULL, compressok || rawok, savedok, &zc->zc_objset_type); if (fromsnap != NULL) dsl_dataset_rele(fromsnap, FTAG); dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } else { zfs_file_t *fp; dmu_send_outparams_t out = {0}; if ((fp = zfs_file_get(zc->zc_cookie)) == NULL) return (SET_ERROR(EBADF)); off = zfs_file_off(fp); out.dso_outfunc = dump_bytes; out.dso_arg = fp; out.dso_dryrun = B_FALSE; error = dmu_send_obj(zc->zc_name, zc->zc_sendobj, zc->zc_fromobj, embedok, large_block_ok, compressok, rawok, savedok, zc->zc_cookie, &off, &out); zfs_file_put(fp); } return (error); } /* * inputs: * zc_name name of snapshot on which to report progress * zc_cookie file descriptor of send stream * * outputs: * zc_cookie number of bytes written in send stream thus far * zc_objset_type logical size of data traversed by send thus far */ static int zfs_ioc_send_progress(zfs_cmd_t *zc) { dsl_pool_t *dp; dsl_dataset_t *ds; dmu_sendstatus_t *dsp = NULL; int error; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } mutex_enter(&ds->ds_sendstream_lock); /* * Iterate over all the send streams currently active on this dataset. * If there's one which matches the specified file descriptor _and_ the * stream was started by the current process, return the progress of * that stream. */ for (dsp = list_head(&ds->ds_sendstreams); dsp != NULL; dsp = list_next(&ds->ds_sendstreams, dsp)) { if (dsp->dss_outfd == zc->zc_cookie && zfs_proc_is_caller(dsp->dss_proc)) break; } if (dsp != NULL) { zc->zc_cookie = atomic_cas_64((volatile uint64_t *)dsp->dss_off, 0, 0); /* This is the closest thing we have to atomic_read_64. */ zc->zc_objset_type = atomic_cas_64(&dsp->dss_blocks, 0, 0); } else { error = SET_ERROR(ENOENT); } mutex_exit(&ds->ds_sendstream_lock); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static int zfs_ioc_inject_fault(zfs_cmd_t *zc) { int id, error; error = zio_inject_fault(zc->zc_name, (int)zc->zc_guid, &id, &zc->zc_inject_record); if (error == 0) zc->zc_guid = (uint64_t)id; return (error); } static int zfs_ioc_clear_fault(zfs_cmd_t *zc) { return (zio_clear_fault((int)zc->zc_guid)); } static int zfs_ioc_inject_list_next(zfs_cmd_t *zc) { int id = (int)zc->zc_guid; int error; error = zio_inject_list_next(&id, zc->zc_name, sizeof (zc->zc_name), &zc->zc_inject_record); zc->zc_guid = id; return (error); } static int zfs_ioc_error_log(zfs_cmd_t *zc) { spa_t *spa; int error; uint64_t count = zc->zc_nvlist_dst_size; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_get_errlog(spa, (void *)(uintptr_t)zc->zc_nvlist_dst, &count); if (error == 0) zc->zc_nvlist_dst_size = count; else zc->zc_nvlist_dst_size = spa_get_errlog_size(spa); spa_close(spa, FTAG); return (error); } static int zfs_ioc_clear(zfs_cmd_t *zc) { spa_t *spa; vdev_t *vd; int error; /* * On zpool clear we also fix up missing slogs */ mutex_enter(&spa_namespace_lock); spa = spa_lookup(zc->zc_name); if (spa == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EIO)); } if (spa_get_log_state(spa) == SPA_LOG_MISSING) { /* we need to let spa_open/spa_load clear the chains */ spa_set_log_state(spa, SPA_LOG_CLEAR); } spa->spa_last_open_failed = 0; mutex_exit(&spa_namespace_lock); if (zc->zc_cookie & ZPOOL_NO_REWIND) { error = spa_open(zc->zc_name, &spa, FTAG); } else { nvlist_t *policy; nvlist_t *config = NULL; if (zc->zc_nvlist_src == 0) return (SET_ERROR(EINVAL)); if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &policy)) == 0) { error = spa_open_rewind(zc->zc_name, &spa, FTAG, policy, &config); if (config != NULL) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; nvlist_free(config); } nvlist_free(policy); } } if (error != 0) return (error); /* * If multihost is enabled, resuming I/O is unsafe as another * host may have imported the pool. */ if (spa_multihost(spa) && spa_suspended(spa)) return (SET_ERROR(EINVAL)); spa_vdev_state_enter(spa, SCL_NONE); if (zc->zc_guid == 0) { vd = NULL; } else { vd = spa_lookup_by_guid(spa, zc->zc_guid, B_TRUE); if (vd == NULL) { error = SET_ERROR(ENODEV); (void) spa_vdev_state_exit(spa, NULL, error); spa_close(spa, FTAG); return (error); } } vdev_clear(spa, vd); (void) spa_vdev_state_exit(spa, spa_suspended(spa) ? NULL : spa->spa_root_vdev, 0); /* * Resume any suspended I/Os. */ if (zio_resume(spa) != 0) error = SET_ERROR(EIO); spa_close(spa, FTAG); return (error); } /* * Reopen all the vdevs associated with the pool. * * innvl: { * "scrub_restart" -> when true and scrub is running, allow to restart * scrub as the side effect of the reopen (boolean). * } * * outnvl is unused */ static const zfs_ioc_key_t zfs_keys_pool_reopen[] = { {"scrub_restart", DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL}, }; static int zfs_ioc_pool_reopen(const char *pool, nvlist_t *innvl, nvlist_t *outnvl) { (void) outnvl; spa_t *spa; int error; boolean_t rc, scrub_restart = B_TRUE; if (innvl) { error = nvlist_lookup_boolean_value(innvl, "scrub_restart", &rc); if (error == 0) scrub_restart = rc; } error = spa_open(pool, &spa, FTAG); if (error != 0) return (error); spa_vdev_state_enter(spa, SCL_NONE); /* * If the scrub_restart flag is B_FALSE and a scrub is already * in progress then set spa_scrub_reopen flag to B_TRUE so that * we don't restart the scrub as a side effect of the reopen. * Otherwise, let vdev_open() decided if a resilver is required. */ spa->spa_scrub_reopen = (!scrub_restart && dsl_scan_scrubbing(spa->spa_dsl_pool)); vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); spa_close(spa, FTAG); return (0); } /* * inputs: * zc_name name of filesystem * * outputs: * zc_string name of conflicting snapshot, if there is one */ static int zfs_ioc_promote(zfs_cmd_t *zc) { dsl_pool_t *dp; dsl_dataset_t *ds, *ods; char origin[ZFS_MAX_DATASET_NAME_LEN]; char *cp; int error; zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0 || strchr(zc->zc_name, '%')) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (!dsl_dir_is_clone(ds->ds_dir)) { dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (SET_ERROR(EINVAL)); } error = dsl_dataset_hold_obj(dp, dsl_dir_phys(ds->ds_dir)->dd_origin_obj, FTAG, &ods); if (error != 0) { dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } dsl_dataset_name(ods, origin); dsl_dataset_rele(ods, FTAG); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); /* * We don't need to unmount *all* the origin fs's snapshots, but * it's easier. */ cp = strchr(origin, '@'); if (cp) *cp = '\0'; (void) dmu_objset_find(origin, zfs_unmount_snap_cb, NULL, DS_FIND_SNAPSHOTS); return (dsl_dataset_promote(zc->zc_name, zc->zc_string)); } /* * Retrieve a single {user|group|project}{used|quota}@... property. * * inputs: * zc_name name of filesystem * zc_objset_type zfs_userquota_prop_t * zc_value domain name (eg. "S-1-234-567-89") * zc_guid RID/UID/GID * * outputs: * zc_cookie property value */ static int zfs_ioc_userspace_one(zfs_cmd_t *zc) { zfsvfs_t *zfsvfs; int error; if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE); if (error != 0) return (error); error = zfs_userspace_one(zfsvfs, zc->zc_objset_type, zc->zc_value, zc->zc_guid, &zc->zc_cookie); zfsvfs_rele(zfsvfs, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_objset_type zfs_userquota_prop_t * zc_nvlist_dst[_size] buffer to fill (not really an nvlist) * * outputs: * zc_nvlist_dst[_size] data buffer (array of zfs_useracct_t) * zc_cookie zap cursor */ static int zfs_ioc_userspace_many(zfs_cmd_t *zc) { zfsvfs_t *zfsvfs; int bufsize = zc->zc_nvlist_dst_size; if (bufsize <= 0) return (SET_ERROR(ENOMEM)); int error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE); if (error != 0) return (error); void *buf = vmem_alloc(bufsize, KM_SLEEP); error = zfs_userspace_many(zfsvfs, zc->zc_objset_type, &zc->zc_cookie, buf, &zc->zc_nvlist_dst_size); if (error == 0) { error = xcopyout(buf, (void *)(uintptr_t)zc->zc_nvlist_dst, zc->zc_nvlist_dst_size); } vmem_free(buf, bufsize); zfsvfs_rele(zfsvfs, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * none */ static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc) { int error = 0; zfsvfs_t *zfsvfs; if (getzfsvfs(zc->zc_name, &zfsvfs) == 0) { if (!dmu_objset_userused_enabled(zfsvfs->z_os)) { /* * If userused is not enabled, it may be because the * objset needs to be closed & reopened (to grow the * objset_phys_t). Suspend/resume the fs will do that. */ dsl_dataset_t *ds, *newds; ds = dmu_objset_ds(zfsvfs->z_os); error = zfs_suspend_fs(zfsvfs); if (error == 0) { dmu_objset_refresh_ownership(ds, &newds, B_TRUE, zfsvfs); error = zfs_resume_fs(zfsvfs, newds); } } if (error == 0) { mutex_enter(&zfsvfs->z_os->os_upgrade_lock); if (zfsvfs->z_os->os_upgrade_id == 0) { /* clear potential error code and retry */ zfsvfs->z_os->os_upgrade_status = 0; mutex_exit(&zfsvfs->z_os->os_upgrade_lock); dsl_pool_config_enter( dmu_objset_pool(zfsvfs->z_os), FTAG); dmu_objset_userspace_upgrade(zfsvfs->z_os); dsl_pool_config_exit( dmu_objset_pool(zfsvfs->z_os), FTAG); } else { mutex_exit(&zfsvfs->z_os->os_upgrade_lock); } taskq_wait_id(zfsvfs->z_os->os_spa->spa_upgrade_taskq, zfsvfs->z_os->os_upgrade_id); error = zfsvfs->z_os->os_upgrade_status; } zfs_vfs_rele(zfsvfs); } else { objset_t *os; /* XXX kind of reading contents without owning */ error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os); if (error != 0) return (error); mutex_enter(&os->os_upgrade_lock); if (os->os_upgrade_id == 0) { /* clear potential error code and retry */ os->os_upgrade_status = 0; mutex_exit(&os->os_upgrade_lock); dmu_objset_userspace_upgrade(os); } else { mutex_exit(&os->os_upgrade_lock); } dsl_pool_rele(dmu_objset_pool(os), FTAG); taskq_wait_id(os->os_spa->spa_upgrade_taskq, os->os_upgrade_id); error = os->os_upgrade_status; dsl_dataset_rele_flags(dmu_objset_ds(os), DS_HOLD_FLAG_DECRYPT, FTAG); } return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * none */ static int zfs_ioc_id_quota_upgrade(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os); if (error != 0) return (error); if (dmu_objset_userobjspace_upgradable(os) || dmu_objset_projectquota_upgradable(os)) { mutex_enter(&os->os_upgrade_lock); if (os->os_upgrade_id == 0) { /* clear potential error code and retry */ os->os_upgrade_status = 0; mutex_exit(&os->os_upgrade_lock); dmu_objset_id_quota_upgrade(os); } else { mutex_exit(&os->os_upgrade_lock); } dsl_pool_rele(dmu_objset_pool(os), FTAG); taskq_wait_id(os->os_spa->spa_upgrade_taskq, os->os_upgrade_id); error = os->os_upgrade_status; } else { dsl_pool_rele(dmu_objset_pool(os), FTAG); } dsl_dataset_rele_flags(dmu_objset_ds(os), DS_HOLD_FLAG_DECRYPT, FTAG); return (error); } static int zfs_ioc_share(zfs_cmd_t *zc) { return (SET_ERROR(ENOSYS)); } /* * inputs: * zc_name name of containing filesystem * zc_obj object # beyond which we want next in-use object # * * outputs: * zc_obj next in-use object # */ static int zfs_ioc_next_obj(zfs_cmd_t *zc) { objset_t *os = NULL; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) return (error); error = dmu_object_next(os, &zc->zc_obj, B_FALSE, 0); dmu_objset_rele(os, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_value prefix name for snapshot * zc_cleanup_fd cleanup-on-exit file descriptor for calling process * * outputs: * zc_value short name of new snapshot */ static int zfs_ioc_tmp_snapshot(zfs_cmd_t *zc) { char *snap_name; char *hold_name; minor_t minor; zfs_file_t *fp = zfs_onexit_fd_hold(zc->zc_cleanup_fd, &minor); if (fp == NULL) return (SET_ERROR(EBADF)); snap_name = kmem_asprintf("%s-%016llx", zc->zc_value, (u_longlong_t)ddi_get_lbolt64()); hold_name = kmem_asprintf("%%%s", zc->zc_value); int error = dsl_dataset_snapshot_tmp(zc->zc_name, snap_name, minor, hold_name); if (error == 0) (void) strlcpy(zc->zc_value, snap_name, sizeof (zc->zc_value)); kmem_strfree(snap_name); kmem_strfree(hold_name); zfs_onexit_fd_rele(fp); return (error); } /* * inputs: * zc_name name of "to" snapshot * zc_value name of "from" snapshot * zc_cookie file descriptor to write diff data on * * outputs: * dmu_diff_record_t's to the file descriptor */ static int zfs_ioc_diff(zfs_cmd_t *zc) { zfs_file_t *fp; offset_t off; int error; if ((fp = zfs_file_get(zc->zc_cookie)) == NULL) return (SET_ERROR(EBADF)); off = zfs_file_off(fp); error = dmu_diff(zc->zc_name, zc->zc_value, fp, &off); zfs_file_put(fp); return (error); } static int zfs_ioc_smb_acl(zfs_cmd_t *zc) { return (SET_ERROR(ENOTSUP)); } /* * innvl: { * "holds" -> { snapname -> holdname (string), ... } * (optional) "cleanup_fd" -> fd (int32) * } * * outnvl: { * snapname -> error value (int32) * ... * } */ static const zfs_ioc_key_t zfs_keys_hold[] = { {"holds", DATA_TYPE_NVLIST, 0}, {"cleanup_fd", DATA_TYPE_INT32, ZK_OPTIONAL}, }; static int zfs_ioc_hold(const char *pool, nvlist_t *args, nvlist_t *errlist) { (void) pool; nvpair_t *pair; nvlist_t *holds; int cleanup_fd = -1; int error; minor_t minor = 0; zfs_file_t *fp = NULL; holds = fnvlist_lookup_nvlist(args, "holds"); /* make sure the user didn't pass us any invalid (empty) tags */ for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { char *htag; error = nvpair_value_string(pair, &htag); if (error != 0) return (SET_ERROR(error)); if (strlen(htag) == 0) return (SET_ERROR(EINVAL)); } if (nvlist_lookup_int32(args, "cleanup_fd", &cleanup_fd) == 0) { fp = zfs_onexit_fd_hold(cleanup_fd, &minor); if (fp == NULL) return (SET_ERROR(EBADF)); } error = dsl_dataset_user_hold(holds, minor, errlist); if (fp != NULL) { ASSERT3U(minor, !=, 0); zfs_onexit_fd_rele(fp); } return (SET_ERROR(error)); } /* * innvl is not used. * * outnvl: { * holdname -> time added (uint64 seconds since epoch) * ... * } */ static const zfs_ioc_key_t zfs_keys_get_holds[] = { /* no nvl keys */ }; static int zfs_ioc_get_holds(const char *snapname, nvlist_t *args, nvlist_t *outnvl) { (void) args; return (dsl_dataset_get_holds(snapname, outnvl)); } /* * innvl: { * snapname -> { holdname, ... } * ... * } * * outnvl: { * snapname -> error value (int32) * ... * } */ static const zfs_ioc_key_t zfs_keys_release[] = { {"...", DATA_TYPE_NVLIST, ZK_WILDCARDLIST}, }; static int zfs_ioc_release(const char *pool, nvlist_t *holds, nvlist_t *errlist) { (void) pool; return (dsl_dataset_user_release(holds, errlist)); } /* * inputs: * zc_guid flags (ZEVENT_NONBLOCK) * zc_cleanup_fd zevent file descriptor * * outputs: * zc_nvlist_dst next nvlist event * zc_cookie dropped events since last get */ static int zfs_ioc_events_next(zfs_cmd_t *zc) { zfs_zevent_t *ze; nvlist_t *event = NULL; minor_t minor; uint64_t dropped = 0; int error; zfs_file_t *fp = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze); if (fp == NULL) return (SET_ERROR(EBADF)); do { error = zfs_zevent_next(ze, &event, &zc->zc_nvlist_dst_size, &dropped); if (event != NULL) { zc->zc_cookie = dropped; error = put_nvlist(zc, event); nvlist_free(event); } if (zc->zc_guid & ZEVENT_NONBLOCK) break; if ((error == 0) || (error != ENOENT)) break; error = zfs_zevent_wait(ze); if (error != 0) break; } while (1); zfs_zevent_fd_rele(fp); return (error); } /* * outputs: * zc_cookie cleared events count */ static int zfs_ioc_events_clear(zfs_cmd_t *zc) { uint_t count; zfs_zevent_drain_all(&count); zc->zc_cookie = count; return (0); } /* * inputs: * zc_guid eid | ZEVENT_SEEK_START | ZEVENT_SEEK_END * zc_cleanup zevent file descriptor */ static int zfs_ioc_events_seek(zfs_cmd_t *zc) { zfs_zevent_t *ze; minor_t minor; int error; zfs_file_t *fp = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze); if (fp == NULL) return (SET_ERROR(EBADF)); error = zfs_zevent_seek(ze, zc->zc_guid); zfs_zevent_fd_rele(fp); return (error); } /* * inputs: * zc_name name of later filesystem or snapshot * zc_value full name of old snapshot or bookmark * * outputs: * zc_cookie space in bytes * zc_objset_type compressed space in bytes * zc_perm_action uncompressed space in bytes */ static int zfs_ioc_space_written(zfs_cmd_t *zc) { int error; dsl_pool_t *dp; dsl_dataset_t *new; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &new); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (strchr(zc->zc_value, '#') != NULL) { zfs_bookmark_phys_t bmp; error = dsl_bookmark_lookup(dp, zc->zc_value, new, &bmp); if (error == 0) { error = dsl_dataset_space_written_bookmark(&bmp, new, &zc->zc_cookie, &zc->zc_objset_type, &zc->zc_perm_action); } } else { dsl_dataset_t *old; error = dsl_dataset_hold(dp, zc->zc_value, FTAG, &old); if (error == 0) { error = dsl_dataset_space_written(old, new, &zc->zc_cookie, &zc->zc_objset_type, &zc->zc_perm_action); dsl_dataset_rele(old, FTAG); } } dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* * innvl: { * "firstsnap" -> snapshot name * } * * outnvl: { * "used" -> space in bytes * "compressed" -> compressed space in bytes * "uncompressed" -> uncompressed space in bytes * } */ static const zfs_ioc_key_t zfs_keys_space_snaps[] = { {"firstsnap", DATA_TYPE_STRING, 0}, }; static int zfs_ioc_space_snaps(const char *lastsnap, nvlist_t *innvl, nvlist_t *outnvl) { int error; dsl_pool_t *dp; dsl_dataset_t *new, *old; char *firstsnap; uint64_t used, comp, uncomp; firstsnap = fnvlist_lookup_string(innvl, "firstsnap"); error = dsl_pool_hold(lastsnap, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, lastsnap, FTAG, &new); if (error == 0 && !new->ds_is_snapshot) { dsl_dataset_rele(new, FTAG); error = SET_ERROR(EINVAL); } if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_hold(dp, firstsnap, FTAG, &old); if (error == 0 && !old->ds_is_snapshot) { dsl_dataset_rele(old, FTAG); error = SET_ERROR(EINVAL); } if (error != 0) { dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_space_wouldfree(old, new, &used, &comp, &uncomp); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); fnvlist_add_uint64(outnvl, "used", used); fnvlist_add_uint64(outnvl, "compressed", comp); fnvlist_add_uint64(outnvl, "uncompressed", uncomp); return (error); } /* * innvl: { * "fd" -> file descriptor to write stream to (int32) * (optional) "fromsnap" -> full snap name to send an incremental from * (optional) "largeblockok" -> (value ignored) * indicates that blocks > 128KB are permitted * (optional) "embedok" -> (value ignored) * presence indicates DRR_WRITE_EMBEDDED records are permitted * (optional) "compressok" -> (value ignored) * presence indicates compressed DRR_WRITE records are permitted * (optional) "rawok" -> (value ignored) * presence indicates raw encrypted records should be used. * (optional) "savedok" -> (value ignored) * presence indicates we should send a partially received snapshot * (optional) "resume_object" and "resume_offset" -> (uint64) * if present, resume send stream from specified object and offset. * (optional) "redactbook" -> (string) * if present, use this bookmark's redaction list to generate a redacted * send stream * } * * outnvl is unused */ static const zfs_ioc_key_t zfs_keys_send_new[] = { {"fd", DATA_TYPE_INT32, 0}, {"fromsnap", DATA_TYPE_STRING, ZK_OPTIONAL}, {"largeblockok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"embedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"compressok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"rawok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"savedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"resume_object", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"resume_offset", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"redactbook", DATA_TYPE_STRING, ZK_OPTIONAL}, }; static int zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { (void) outnvl; int error; offset_t off; char *fromname = NULL; int fd; zfs_file_t *fp; boolean_t largeblockok; boolean_t embedok; boolean_t compressok; boolean_t rawok; boolean_t savedok; uint64_t resumeobj = 0; uint64_t resumeoff = 0; char *redactbook = NULL; fd = fnvlist_lookup_int32(innvl, "fd"); (void) nvlist_lookup_string(innvl, "fromsnap", &fromname); largeblockok = nvlist_exists(innvl, "largeblockok"); embedok = nvlist_exists(innvl, "embedok"); compressok = nvlist_exists(innvl, "compressok"); rawok = nvlist_exists(innvl, "rawok"); savedok = nvlist_exists(innvl, "savedok"); (void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj); (void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff); (void) nvlist_lookup_string(innvl, "redactbook", &redactbook); if ((fp = zfs_file_get(fd)) == NULL) return (SET_ERROR(EBADF)); off = zfs_file_off(fp); dmu_send_outparams_t out = {0}; out.dso_outfunc = dump_bytes; out.dso_arg = fp; out.dso_dryrun = B_FALSE; error = dmu_send(snapname, fromname, embedok, largeblockok, compressok, rawok, savedok, resumeobj, resumeoff, redactbook, fd, &off, &out); zfs_file_put(fp); return (error); } static int send_space_sum(objset_t *os, void *buf, int len, void *arg) { (void) os, (void) buf; uint64_t *size = arg; *size += len; return (0); } /* * Determine approximately how large a zfs send stream will be -- the number * of bytes that will be written to the fd supplied to zfs_ioc_send_new(). * * innvl: { * (optional) "from" -> full snap or bookmark name to send an incremental * from * (optional) "largeblockok" -> (value ignored) * indicates that blocks > 128KB are permitted * (optional) "embedok" -> (value ignored) * presence indicates DRR_WRITE_EMBEDDED records are permitted * (optional) "compressok" -> (value ignored) * presence indicates compressed DRR_WRITE records are permitted * (optional) "rawok" -> (value ignored) * presence indicates raw encrypted records should be used. * (optional) "resume_object" and "resume_offset" -> (uint64) * if present, resume send stream from specified object and offset. * (optional) "fd" -> file descriptor to use as a cookie for progress * tracking (int32) * } * * outnvl: { * "space" -> bytes of space (uint64) * } */ static const zfs_ioc_key_t zfs_keys_send_space[] = { {"from", DATA_TYPE_STRING, ZK_OPTIONAL}, {"fromsnap", DATA_TYPE_STRING, ZK_OPTIONAL}, {"largeblockok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"embedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"compressok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"rawok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, {"fd", DATA_TYPE_INT32, ZK_OPTIONAL}, {"redactbook", DATA_TYPE_STRING, ZK_OPTIONAL}, {"resume_object", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"resume_offset", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"bytes", DATA_TYPE_UINT64, ZK_OPTIONAL}, }; static int zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { dsl_pool_t *dp; dsl_dataset_t *tosnap; dsl_dataset_t *fromsnap = NULL; int error; char *fromname = NULL; char *redactlist_book = NULL; boolean_t largeblockok; boolean_t embedok; boolean_t compressok; boolean_t rawok; boolean_t savedok; uint64_t space = 0; boolean_t full_estimate = B_FALSE; uint64_t resumeobj = 0; uint64_t resumeoff = 0; uint64_t resume_bytes = 0; int32_t fd = -1; zfs_bookmark_phys_t zbm = {0}; error = dsl_pool_hold(snapname, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, snapname, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } (void) nvlist_lookup_int32(innvl, "fd", &fd); largeblockok = nvlist_exists(innvl, "largeblockok"); embedok = nvlist_exists(innvl, "embedok"); compressok = nvlist_exists(innvl, "compressok"); rawok = nvlist_exists(innvl, "rawok"); savedok = nvlist_exists(innvl, "savedok"); boolean_t from = (nvlist_lookup_string(innvl, "from", &fromname) == 0); boolean_t altbook = (nvlist_lookup_string(innvl, "redactbook", &redactlist_book) == 0); (void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj); (void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff); (void) nvlist_lookup_uint64(innvl, "bytes", &resume_bytes); if (altbook) { full_estimate = B_TRUE; } else if (from) { if (strchr(fromname, '#')) { error = dsl_bookmark_lookup(dp, fromname, tosnap, &zbm); /* * dsl_bookmark_lookup() will fail with EXDEV if * the from-bookmark and tosnap are at the same txg. * However, it's valid to do a send (and therefore, * a send estimate) from and to the same time point, * if the bookmark is redacted (the incremental send * can change what's redacted on the target). In * this case, dsl_bookmark_lookup() fills in zbm * but returns EXDEV. Ignore this error. */ if (error == EXDEV && zbm.zbm_redaction_obj != 0 && zbm.zbm_guid == dsl_dataset_phys(tosnap)->ds_guid) error = 0; if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } if (zbm.zbm_redaction_obj != 0 || !(zbm.zbm_flags & ZBM_FLAG_HAS_FBN)) { full_estimate = B_TRUE; } } else if (strchr(fromname, '@')) { error = dsl_dataset_hold(dp, fromname, FTAG, &fromsnap); if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } if (!dsl_dataset_is_before(tosnap, fromsnap, 0)) { full_estimate = B_TRUE; dsl_dataset_rele(fromsnap, FTAG); } } else { /* * from is not properly formatted as a snapshot or * bookmark */ dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (SET_ERROR(EINVAL)); } } if (full_estimate) { dmu_send_outparams_t out = {0}; offset_t off = 0; out.dso_outfunc = send_space_sum; out.dso_arg = &space; out.dso_dryrun = B_TRUE; /* * We have to release these holds so dmu_send can take them. It * will do all the error checking we need. */ dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); error = dmu_send(snapname, fromname, embedok, largeblockok, compressok, rawok, savedok, resumeobj, resumeoff, redactlist_book, fd, &off, &out); } else { error = dmu_send_estimate_fast(tosnap, fromsnap, (from && strchr(fromname, '#') != NULL ? &zbm : NULL), compressok || rawok, savedok, &space); space -= resume_bytes; if (fromsnap != NULL) dsl_dataset_rele(fromsnap, FTAG); dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } fnvlist_add_uint64(outnvl, "space", space); return (error); } /* * Sync the currently open TXG to disk for the specified pool. * This is somewhat similar to 'zfs_sync()'. * For cases that do not result in error this ioctl will wait for * the currently open TXG to commit before returning back to the caller. * * innvl: { * "force" -> when true, force uberblock update even if there is no dirty data. * In addition this will cause the vdev configuration to be written * out including updating the zpool cache file. (boolean_t) * } * * onvl is unused */ static const zfs_ioc_key_t zfs_keys_pool_sync[] = { {"force", DATA_TYPE_BOOLEAN_VALUE, 0}, }; static int zfs_ioc_pool_sync(const char *pool, nvlist_t *innvl, nvlist_t *onvl) { (void) onvl; int err; boolean_t rc, force = B_FALSE; spa_t *spa; if ((err = spa_open(pool, &spa, FTAG)) != 0) return (err); if (innvl) { err = nvlist_lookup_boolean_value(innvl, "force", &rc); if (err == 0) force = rc; } if (force) { spa_config_enter(spa, SCL_CONFIG, FTAG, RW_WRITER); vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); } txg_wait_synced(spa_get_dsl(spa), 0); spa_close(spa, FTAG); return (0); } /* * Load a user's wrapping key into the kernel. * innvl: { * "hidden_args" -> { "wkeydata" -> value } * raw uint8_t array of encryption wrapping key data (32 bytes) * (optional) "noop" -> (value ignored) * presence indicated key should only be verified, not loaded * } */ static const zfs_ioc_key_t zfs_keys_load_key[] = { {"hidden_args", DATA_TYPE_NVLIST, 0}, {"noop", DATA_TYPE_BOOLEAN, ZK_OPTIONAL}, }; static int zfs_ioc_load_key(const char *dsname, nvlist_t *innvl, nvlist_t *outnvl) { (void) outnvl; int ret; dsl_crypto_params_t *dcp = NULL; nvlist_t *hidden_args; boolean_t noop = nvlist_exists(innvl, "noop"); if (strchr(dsname, '@') != NULL || strchr(dsname, '%') != NULL) { ret = SET_ERROR(EINVAL); goto error; } hidden_args = fnvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS); ret = dsl_crypto_params_create_nvlist(DCP_CMD_NONE, NULL, hidden_args, &dcp); if (ret != 0) goto error; ret = spa_keystore_load_wkey(dsname, dcp, noop); if (ret != 0) goto error; dsl_crypto_params_free(dcp, noop); return (0); error: dsl_crypto_params_free(dcp, B_TRUE); return (ret); } /* * Unload a user's wrapping key from the kernel. * Both innvl and outnvl are unused. */ static const zfs_ioc_key_t zfs_keys_unload_key[] = { /* no nvl keys */ }; static int zfs_ioc_unload_key(const char *dsname, nvlist_t *innvl, nvlist_t *outnvl) { (void) innvl, (void) outnvl; int ret = 0; if (strchr(dsname, '@') != NULL || strchr(dsname, '%') != NULL) { ret = (SET_ERROR(EINVAL)); goto out; } ret = spa_keystore_unload_wkey(dsname); if (ret != 0) goto out; out: return (ret); } /* * Changes a user's wrapping key used to decrypt a dataset. The keyformat, * keylocation, pbkdf2salt, and pbkdf2iters properties can also be specified * here to change how the key is derived in userspace. * * innvl: { * "hidden_args" (optional) -> { "wkeydata" -> value } * raw uint8_t array of new encryption wrapping key data (32 bytes) * "props" (optional) -> { prop -> value } * } * * outnvl is unused */ static const zfs_ioc_key_t zfs_keys_change_key[] = { {"crypt_cmd", DATA_TYPE_UINT64, ZK_OPTIONAL}, {"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL}, {"props", DATA_TYPE_NVLIST, ZK_OPTIONAL}, }; static int zfs_ioc_change_key(const char *dsname, nvlist_t *innvl, nvlist_t *outnvl) { (void) outnvl; int ret; uint64_t cmd = DCP_CMD_NONE; dsl_crypto_params_t *dcp = NULL; nvlist_t *args = NULL, *hidden_args = NULL; if (strchr(dsname, '@') != NULL || strchr(dsname, '%') != NULL) { ret = (SET_ERROR(EINVAL)); goto error; } (void) nvlist_lookup_uint64(innvl, "crypt_cmd", &cmd); (void) nvlist_lookup_nvlist(innvl, "props", &args); (void) nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args); ret = dsl_crypto_params_create_nvlist(cmd, args, hidden_args, &dcp); if (ret != 0) goto error; ret = spa_keystore_change_key(dsname, dcp); if (ret != 0) goto error; dsl_crypto_params_free(dcp, B_FALSE); return (0); error: dsl_crypto_params_free(dcp, B_TRUE); return (ret); } static zfs_ioc_vec_t zfs_ioc_vec[ZFS_IOC_LAST - ZFS_IOC_FIRST]; static void zfs_ioctl_register_legacy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); vec->zvec_legacy_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_allow_log = log_history; vec->zvec_pool_check = pool_check; } /* * See the block comment at the beginning of this file for details on * each argument to this function. */ void zfs_ioctl_register(const char *name, zfs_ioc_t ioc, zfs_ioc_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, zfs_ioc_poolcheck_t pool_check, boolean_t smush_outnvlist, boolean_t allow_log, const zfs_ioc_key_t *nvl_keys, size_t num_keys) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); /* if we are logging, the name must be valid */ ASSERT(!allow_log || namecheck != NO_NAME); vec->zvec_name = name; vec->zvec_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_pool_check = pool_check; vec->zvec_smush_outnvlist = smush_outnvlist; vec->zvec_allow_log = allow_log; vec->zvec_nvl_keys = nvl_keys; vec->zvec_nvl_key_count = num_keys; } static void zfs_ioctl_register_pool(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, POOL_NAME, log_history, pool_check); } void zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, pool_check); } static void zfs_ioctl_register_pool_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_legacy(ioc, func, zfs_secpolicy_config, POOL_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_register_pool_meta(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, NO_NAME, B_FALSE, POOL_CHECK_NONE); } static void zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED); } static void zfs_ioctl_register_dataset_read(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_dataset_read_secpolicy(ioc, func, zfs_secpolicy_read); } static void zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_init(void) { zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT, zfs_ioc_snapshot, zfs_secpolicy_snapshot, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_snapshot, ARRAY_SIZE(zfs_keys_snapshot)); zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY, zfs_ioc_log_history, zfs_secpolicy_log_history, NO_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE, zfs_keys_log_history, ARRAY_SIZE(zfs_keys_log_history)); zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS, zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_space_snaps, ARRAY_SIZE(zfs_keys_space_snaps)); zfs_ioctl_register("send", ZFS_IOC_SEND_NEW, zfs_ioc_send_new, zfs_secpolicy_send_new, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_send_new, ARRAY_SIZE(zfs_keys_send_new)); zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE, zfs_ioc_send_space, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_send_space, ARRAY_SIZE(zfs_keys_send_space)); zfs_ioctl_register("create", ZFS_IOC_CREATE, zfs_ioc_create, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_create, ARRAY_SIZE(zfs_keys_create)); zfs_ioctl_register("clone", ZFS_IOC_CLONE, zfs_ioc_clone, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_clone, ARRAY_SIZE(zfs_keys_clone)); zfs_ioctl_register("remap", ZFS_IOC_REMAP, zfs_ioc_remap, zfs_secpolicy_none, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE, zfs_keys_remap, ARRAY_SIZE(zfs_keys_remap)); zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS, zfs_ioc_destroy_snaps, zfs_secpolicy_destroy_snaps, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_destroy_snaps, ARRAY_SIZE(zfs_keys_destroy_snaps)); zfs_ioctl_register("hold", ZFS_IOC_HOLD, zfs_ioc_hold, zfs_secpolicy_hold, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_hold, ARRAY_SIZE(zfs_keys_hold)); zfs_ioctl_register("release", ZFS_IOC_RELEASE, zfs_ioc_release, zfs_secpolicy_release, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_release, ARRAY_SIZE(zfs_keys_release)); zfs_ioctl_register("get_holds", ZFS_IOC_GET_HOLDS, zfs_ioc_get_holds, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_get_holds, ARRAY_SIZE(zfs_keys_get_holds)); zfs_ioctl_register("rollback", ZFS_IOC_ROLLBACK, zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE, zfs_keys_rollback, ARRAY_SIZE(zfs_keys_rollback)); zfs_ioctl_register("bookmark", ZFS_IOC_BOOKMARK, zfs_ioc_bookmark, zfs_secpolicy_bookmark, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_bookmark, ARRAY_SIZE(zfs_keys_bookmark)); zfs_ioctl_register("get_bookmarks", ZFS_IOC_GET_BOOKMARKS, zfs_ioc_get_bookmarks, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_get_bookmarks, ARRAY_SIZE(zfs_keys_get_bookmarks)); zfs_ioctl_register("get_bookmark_props", ZFS_IOC_GET_BOOKMARK_PROPS, zfs_ioc_get_bookmark_props, zfs_secpolicy_read, ENTITY_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_get_bookmark_props, ARRAY_SIZE(zfs_keys_get_bookmark_props)); zfs_ioctl_register("destroy_bookmarks", ZFS_IOC_DESTROY_BOOKMARKS, zfs_ioc_destroy_bookmarks, zfs_secpolicy_destroy_bookmarks, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_destroy_bookmarks, ARRAY_SIZE(zfs_keys_destroy_bookmarks)); zfs_ioctl_register("receive", ZFS_IOC_RECV_NEW, zfs_ioc_recv_new, zfs_secpolicy_recv, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_recv_new, ARRAY_SIZE(zfs_keys_recv_new)); zfs_ioctl_register("load-key", ZFS_IOC_LOAD_KEY, zfs_ioc_load_key, zfs_secpolicy_load_key, DATASET_NAME, POOL_CHECK_SUSPENDED, B_TRUE, B_TRUE, zfs_keys_load_key, ARRAY_SIZE(zfs_keys_load_key)); zfs_ioctl_register("unload-key", ZFS_IOC_UNLOAD_KEY, zfs_ioc_unload_key, zfs_secpolicy_load_key, DATASET_NAME, POOL_CHECK_SUSPENDED, B_TRUE, B_TRUE, zfs_keys_unload_key, ARRAY_SIZE(zfs_keys_unload_key)); zfs_ioctl_register("change-key", ZFS_IOC_CHANGE_KEY, zfs_ioc_change_key, zfs_secpolicy_change_key, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_change_key, ARRAY_SIZE(zfs_keys_change_key)); zfs_ioctl_register("sync", ZFS_IOC_POOL_SYNC, zfs_ioc_pool_sync, zfs_secpolicy_none, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE, zfs_keys_pool_sync, ARRAY_SIZE(zfs_keys_pool_sync)); zfs_ioctl_register("reopen", ZFS_IOC_POOL_REOPEN, zfs_ioc_pool_reopen, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED, B_TRUE, B_TRUE, zfs_keys_pool_reopen, ARRAY_SIZE(zfs_keys_pool_reopen)); zfs_ioctl_register("channel_program", ZFS_IOC_CHANNEL_PROGRAM, zfs_ioc_channel_program, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_channel_program, ARRAY_SIZE(zfs_keys_channel_program)); zfs_ioctl_register("redact", ZFS_IOC_REDACT, zfs_ioc_redact, zfs_secpolicy_config, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_redact, ARRAY_SIZE(zfs_keys_redact)); zfs_ioctl_register("zpool_checkpoint", ZFS_IOC_POOL_CHECKPOINT, zfs_ioc_pool_checkpoint, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_pool_checkpoint, ARRAY_SIZE(zfs_keys_pool_checkpoint)); zfs_ioctl_register("zpool_discard_checkpoint", ZFS_IOC_POOL_DISCARD_CHECKPOINT, zfs_ioc_pool_discard_checkpoint, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_pool_discard_checkpoint, ARRAY_SIZE(zfs_keys_pool_discard_checkpoint)); zfs_ioctl_register("initialize", ZFS_IOC_POOL_INITIALIZE, zfs_ioc_pool_initialize, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_pool_initialize, ARRAY_SIZE(zfs_keys_pool_initialize)); zfs_ioctl_register("trim", ZFS_IOC_POOL_TRIM, zfs_ioc_pool_trim, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE, zfs_keys_pool_trim, ARRAY_SIZE(zfs_keys_pool_trim)); zfs_ioctl_register("wait", ZFS_IOC_WAIT, zfs_ioc_wait, zfs_secpolicy_none, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE, zfs_keys_pool_wait, ARRAY_SIZE(zfs_keys_pool_wait)); zfs_ioctl_register("wait_fs", ZFS_IOC_WAIT_FS, zfs_ioc_wait_fs, zfs_secpolicy_none, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE, zfs_keys_fs_wait, ARRAY_SIZE(zfs_keys_fs_wait)); zfs_ioctl_register("set_bootenv", ZFS_IOC_SET_BOOTENV, zfs_ioc_set_bootenv, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE, zfs_keys_set_bootenv, ARRAY_SIZE(zfs_keys_set_bootenv)); zfs_ioctl_register("get_bootenv", ZFS_IOC_GET_BOOTENV, zfs_ioc_get_bootenv, zfs_secpolicy_none, POOL_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_TRUE, zfs_keys_get_bootenv, ARRAY_SIZE(zfs_keys_get_bootenv)); zfs_ioctl_register("zpool_vdev_get_props", ZFS_IOC_VDEV_GET_PROPS, zfs_ioc_vdev_get_props, zfs_secpolicy_read, POOL_NAME, POOL_CHECK_NONE, B_FALSE, B_FALSE, zfs_keys_vdev_get_props, ARRAY_SIZE(zfs_keys_vdev_get_props)); zfs_ioctl_register("zpool_vdev_set_props", ZFS_IOC_VDEV_SET_PROPS, zfs_ioc_vdev_set_props, zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE, zfs_keys_vdev_set_props, ARRAY_SIZE(zfs_keys_vdev_set_props)); /* IOCTLS that use the legacy function signature */ zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_READONLY); zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE, zfs_ioc_pool_create, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN, zfs_ioc_pool_scan); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE, zfs_ioc_pool_upgrade); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD, zfs_ioc_vdev_add); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE, zfs_ioc_vdev_remove); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE, zfs_ioc_vdev_set_state); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH, zfs_ioc_vdev_attach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH, zfs_ioc_vdev_detach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH, zfs_ioc_vdev_setpath); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU, zfs_ioc_vdev_setfru); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS, zfs_ioc_pool_set_props); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT, zfs_ioc_vdev_split); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID, zfs_ioc_pool_reguid); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS, zfs_ioc_pool_configs, zfs_secpolicy_none); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT, zfs_ioc_pool_tryimport, zfs_secpolicy_config); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT, zfs_ioc_inject_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT, zfs_ioc_clear_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT, zfs_ioc_inject_list_next, zfs_secpolicy_inject); /* * pool destroy, and export don't log the history as part of * zfsdev_ioctl, but rather zfs_ioc_pool_export * does the logging of those commands. */ zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY, zfs_ioc_pool_destroy, zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT, zfs_ioc_pool_export, zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS, zfs_ioc_pool_stats, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS, zfs_ioc_pool_get_props, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG, zfs_ioc_error_log, zfs_secpolicy_inject, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME, zfs_ioc_dsobj_to_dsname, zfs_secpolicy_diff, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY, zfs_ioc_pool_get_history, zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT, zfs_ioc_pool_import, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_CLEAR, zfs_ioc_clear, zfs_secpolicy_config, B_TRUE, POOL_CHECK_READONLY); zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN, zfs_ioc_space_written); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS, zfs_ioc_objset_recvd_props); zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ, zfs_ioc_next_obj); zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL, zfs_ioc_get_fsacl); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS, zfs_ioc_objset_stats); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS, zfs_ioc_objset_zplprops); zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT, zfs_ioc_dataset_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT, zfs_ioc_snapshot_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS, zfs_ioc_send_progress); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF, zfs_ioc_diff, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS, zfs_ioc_obj_to_stats, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH, zfs_ioc_obj_to_path, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE, zfs_ioc_userspace_one, zfs_secpolicy_userspace_one); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY, zfs_ioc_userspace_many, zfs_secpolicy_userspace_many); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND, zfs_ioc_send, zfs_secpolicy_send); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP, zfs_ioc_set_prop, zfs_secpolicy_none); zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY, zfs_ioc_destroy, zfs_secpolicy_destroy); zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME, zfs_ioc_rename, zfs_secpolicy_rename); zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV, zfs_ioc_recv, zfs_secpolicy_recv); zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE, zfs_ioc_promote, zfs_secpolicy_promote); zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP, zfs_ioc_inherit_prop, zfs_secpolicy_inherit_prop); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL, zfs_ioc_set_fsacl, zfs_secpolicy_set_fsacl); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE, zfs_ioc_share, zfs_secpolicy_share, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL, zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE, zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT, zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_NEXT, zfs_ioc_events_next, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_CLEAR, zfs_ioc_events_clear, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_SEEK, zfs_ioc_events_seek, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_init_os(); } /* * Verify that for non-legacy ioctls the input nvlist * pairs match against the expected input. * * Possible errors are: * ZFS_ERR_IOC_ARG_UNAVAIL An unrecognized nvpair was encountered * ZFS_ERR_IOC_ARG_REQUIRED A required nvpair is missing * ZFS_ERR_IOC_ARG_BADTYPE Invalid type for nvpair */ static int zfs_check_input_nvpairs(nvlist_t *innvl, const zfs_ioc_vec_t *vec) { const zfs_ioc_key_t *nvl_keys = vec->zvec_nvl_keys; boolean_t required_keys_found = B_FALSE; /* * examine each input pair */ for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *name = nvpair_name(pair); data_type_t type = nvpair_type(pair); boolean_t identified = B_FALSE; /* * check pair against the documented names and type */ for (int k = 0; k < vec->zvec_nvl_key_count; k++) { /* if not a wild card name, check for an exact match */ if ((nvl_keys[k].zkey_flags & ZK_WILDCARDLIST) == 0 && strcmp(nvl_keys[k].zkey_name, name) != 0) continue; identified = B_TRUE; if (nvl_keys[k].zkey_type != DATA_TYPE_ANY && nvl_keys[k].zkey_type != type) { return (SET_ERROR(ZFS_ERR_IOC_ARG_BADTYPE)); } if (nvl_keys[k].zkey_flags & ZK_OPTIONAL) continue; required_keys_found = B_TRUE; break; } /* allow an 'optional' key, everything else is invalid */ if (!identified && (strcmp(name, "optional") != 0 || type != DATA_TYPE_NVLIST)) { return (SET_ERROR(ZFS_ERR_IOC_ARG_UNAVAIL)); } } /* verify that all required keys were found */ for (int k = 0; k < vec->zvec_nvl_key_count; k++) { if (nvl_keys[k].zkey_flags & ZK_OPTIONAL) continue; if (nvl_keys[k].zkey_flags & ZK_WILDCARDLIST) { /* at least one non-optional key is expected here */ if (!required_keys_found) return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED)); continue; } if (!nvlist_exists(innvl, nvl_keys[k].zkey_name)) return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED)); } return (0); } static int pool_status_check(const char *name, zfs_ioc_namecheck_t type, zfs_ioc_poolcheck_t check) { spa_t *spa; int error; ASSERT(type == POOL_NAME || type == DATASET_NAME || type == ENTITY_NAME); if (check & POOL_CHECK_NONE) return (0); error = spa_open(name, &spa, FTAG); if (error == 0) { if ((check & POOL_CHECK_SUSPENDED) && spa_suspended(spa)) error = SET_ERROR(EAGAIN); else if ((check & POOL_CHECK_READONLY) && !spa_writeable(spa)) error = SET_ERROR(EROFS); spa_close(spa, FTAG); } return (error); } int zfsdev_getminor(zfs_file_t *fp, minor_t *minorp) { zfsdev_state_t *zs, *fpd; ASSERT(!MUTEX_HELD(&zfsdev_state_lock)); fpd = zfs_file_private(fp); if (fpd == NULL) return (SET_ERROR(EBADF)); mutex_enter(&zfsdev_state_lock); for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zs->zs_minor == -1) continue; if (fpd == zs) { *minorp = fpd->zs_minor; mutex_exit(&zfsdev_state_lock); return (0); } } mutex_exit(&zfsdev_state_lock); return (SET_ERROR(EBADF)); } void * zfsdev_get_state(minor_t minor, enum zfsdev_state_type which) { zfsdev_state_t *zs; for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zs->zs_minor == minor) { membar_consumer(); switch (which) { case ZST_ONEXIT: return (zs->zs_onexit); case ZST_ZEVENT: return (zs->zs_zevent); case ZST_ALL: return (zs); } } } return (NULL); } /* * Find a free minor number. The zfsdev_state_list is expected to * be short since it is only a list of currently open file handles. */ static minor_t zfsdev_minor_alloc(void) { static minor_t last_minor = 0; minor_t m; ASSERT(MUTEX_HELD(&zfsdev_state_lock)); for (m = last_minor + 1; m != last_minor; m++) { if (m > ZFSDEV_MAX_MINOR) m = 1; if (zfsdev_get_state(m, ZST_ALL) == NULL) { last_minor = m; return (m); } } return (0); } int zfsdev_state_init(void *priv) { zfsdev_state_t *zs, *zsprev = NULL; minor_t minor; boolean_t newzs = B_FALSE; ASSERT(MUTEX_HELD(&zfsdev_state_lock)); minor = zfsdev_minor_alloc(); if (minor == 0) return (SET_ERROR(ENXIO)); for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) { if (zs->zs_minor == -1) break; zsprev = zs; } if (!zs) { zs = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP); newzs = B_TRUE; } zfsdev_private_set_state(priv, zs); zfs_onexit_init((zfs_onexit_t **)&zs->zs_onexit); zfs_zevent_init((zfs_zevent_t **)&zs->zs_zevent); /* * In order to provide for lock-free concurrent read access * to the minor list in zfsdev_get_state(), new entries * must be completely written before linking them into the * list whereas existing entries are already linked; the last * operation must be updating zs_minor (from -1 to the new * value). */ if (newzs) { zs->zs_minor = minor; membar_producer(); zsprev->zs_next = zs; } else { membar_producer(); zs->zs_minor = minor; } return (0); } void zfsdev_state_destroy(void *priv) { zfsdev_state_t *zs = zfsdev_private_get_state(priv); ASSERT(zs != NULL); ASSERT3S(zs->zs_minor, >, 0); /* * The last reference to this zfsdev file descriptor is being dropped. * We don't have to worry about lookup grabbing this state object, and * zfsdev_state_init() will not try to reuse this object until it is * invalidated by setting zs_minor to -1. Invalidation must be done * last, with a memory barrier to ensure ordering. This lets us avoid * taking the global zfsdev state lock around destruction. */ zfs_onexit_destroy(zs->zs_onexit); zfs_zevent_destroy(zs->zs_zevent); zs->zs_onexit = NULL; zs->zs_zevent = NULL; membar_producer(); zs->zs_minor = -1; } long zfsdev_ioctl_common(uint_t vecnum, zfs_cmd_t *zc, int flag) { int error, cmd; const zfs_ioc_vec_t *vec; char *saved_poolname = NULL; uint64_t max_nvlist_src_size; size_t saved_poolname_len = 0; nvlist_t *innvl = NULL; fstrans_cookie_t cookie; hrtime_t start_time = gethrtime(); cmd = vecnum; error = 0; if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0])) return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL)); vec = &zfs_ioc_vec[vecnum]; /* * The registered ioctl list may be sparse, verify that either * a normal or legacy handler are registered. */ if (vec->zvec_func == NULL && vec->zvec_legacy_func == NULL) return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL)); zc->zc_iflags = flag & FKIOCTL; max_nvlist_src_size = zfs_max_nvlist_src_size_os(); if (zc->zc_nvlist_src_size > max_nvlist_src_size) { /* * Make sure the user doesn't pass in an insane value for * zc_nvlist_src_size. We have to check, since we will end * up allocating that much memory inside of get_nvlist(). This * prevents a nefarious user from allocating tons of kernel * memory. * * Also, we return EINVAL instead of ENOMEM here. The reason * being that returning ENOMEM from an ioctl() has a special * connotation; that the user's size value is too small and * needs to be expanded to hold the nvlist. See * zcmd_expand_dst_nvlist() for details. */ error = SET_ERROR(EINVAL); /* User's size too big */ } else if (zc->zc_nvlist_src_size != 0) { error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &innvl); if (error != 0) goto out; } /* * Ensure that all pool/dataset names are valid before we pass down to * the lower layers. */ zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; switch (vec->zvec_namecheck) { case POOL_NAME: if (pool_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case DATASET_NAME: if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case ENTITY_NAME: if (entity_namecheck(zc->zc_name, NULL, NULL) != 0) { error = SET_ERROR(EINVAL); } else { error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); } break; case NO_NAME: break; } /* * Ensure that all input pairs are valid before we pass them down * to the lower layers. * * The vectored functions can use fnvlist_lookup_{type} for any * required pairs since zfs_check_input_nvpairs() confirmed that * they exist and are of the correct type. */ if (error == 0 && vec->zvec_func != NULL) { error = zfs_check_input_nvpairs(innvl, vec); if (error != 0) goto out; } if (error == 0) { cookie = spl_fstrans_mark(); error = vec->zvec_secpolicy(zc, innvl, CRED()); spl_fstrans_unmark(cookie); } if (error != 0) goto out; /* legacy ioctls can modify zc_name */ /* * Can't use kmem_strdup() as we might truncate the string and * kmem_strfree() would then free with incorrect size. */ saved_poolname_len = strlen(zc->zc_name) + 1; saved_poolname = kmem_alloc(saved_poolname_len, KM_SLEEP); strlcpy(saved_poolname, zc->zc_name, saved_poolname_len); saved_poolname[strcspn(saved_poolname, "/@#")] = '\0'; if (vec->zvec_func != NULL) { nvlist_t *outnvl; int puterror = 0; spa_t *spa; nvlist_t *lognv = NULL; ASSERT(vec->zvec_legacy_func == NULL); /* * Add the innvl to the lognv before calling the func, * in case the func changes the innvl. */ if (vec->zvec_allow_log) { lognv = fnvlist_alloc(); fnvlist_add_string(lognv, ZPOOL_HIST_IOCTL, vec->zvec_name); if (!nvlist_empty(innvl)) { fnvlist_add_nvlist(lognv, ZPOOL_HIST_INPUT_NVL, innvl); } } outnvl = fnvlist_alloc(); cookie = spl_fstrans_mark(); error = vec->zvec_func(zc->zc_name, innvl, outnvl); spl_fstrans_unmark(cookie); /* * Some commands can partially execute, modify state, and still * return an error. In these cases, attempt to record what * was modified. */ if ((error == 0 || (cmd == ZFS_IOC_CHANNEL_PROGRAM && error != EINVAL)) && vec->zvec_allow_log && spa_open(zc->zc_name, &spa, FTAG) == 0) { if (!nvlist_empty(outnvl)) { size_t out_size = fnvlist_size(outnvl); if (out_size > zfs_history_output_max) { fnvlist_add_int64(lognv, ZPOOL_HIST_OUTPUT_SIZE, out_size); } else { fnvlist_add_nvlist(lognv, ZPOOL_HIST_OUTPUT_NVL, outnvl); } } if (error != 0) { fnvlist_add_int64(lognv, ZPOOL_HIST_ERRNO, error); } fnvlist_add_int64(lognv, ZPOOL_HIST_ELAPSED_NS, gethrtime() - start_time); (void) spa_history_log_nvl(spa, lognv); spa_close(spa, FTAG); } fnvlist_free(lognv); if (!nvlist_empty(outnvl) || zc->zc_nvlist_dst_size != 0) { int smusherror = 0; if (vec->zvec_smush_outnvlist) { smusherror = nvlist_smush(outnvl, zc->zc_nvlist_dst_size); } if (smusherror == 0) puterror = put_nvlist(zc, outnvl); } if (puterror != 0) error = puterror; nvlist_free(outnvl); } else { cookie = spl_fstrans_mark(); error = vec->zvec_legacy_func(zc); spl_fstrans_unmark(cookie); } out: nvlist_free(innvl); if (error == 0 && vec->zvec_allow_log) { char *s = tsd_get(zfs_allow_log_key); if (s != NULL) kmem_strfree(s); (void) tsd_set(zfs_allow_log_key, kmem_strdup(saved_poolname)); } if (saved_poolname != NULL) kmem_free(saved_poolname, saved_poolname_len); return (error); } int zfs_kmod_init(void) { int error; if ((error = zvol_init()) != 0) return (error); spa_init(SPA_MODE_READ | SPA_MODE_WRITE); zfs_init(); zfs_ioctl_init(); mutex_init(&zfsdev_state_lock, NULL, MUTEX_DEFAULT, NULL); zfsdev_state_list = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP); zfsdev_state_list->zs_minor = -1; if ((error = zfsdev_attach()) != 0) goto out; tsd_create(&zfs_fsyncer_key, NULL); tsd_create(&rrw_tsd_key, rrw_tsd_destroy); tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy); return (0); out: zfs_fini(); spa_fini(); zvol_fini(); return (error); } void zfs_kmod_fini(void) { zfsdev_state_t *zs, *zsnext = NULL; zfsdev_detach(); mutex_destroy(&zfsdev_state_lock); for (zs = zfsdev_state_list; zs != NULL; zs = zsnext) { zsnext = zs->zs_next; if (zs->zs_onexit) zfs_onexit_destroy(zs->zs_onexit); if (zs->zs_zevent) zfs_zevent_destroy(zs->zs_zevent); kmem_free(zs, sizeof (zfsdev_state_t)); } zfs_ereport_taskq_fini(); /* run before zfs_fini() on Linux */ zfs_fini(); spa_fini(); zvol_fini(); tsd_destroy(&zfs_fsyncer_key); tsd_destroy(&rrw_tsd_key); tsd_destroy(&zfs_allow_log_key); } ZFS_MODULE_PARAM(zfs, zfs_, max_nvlist_src_size, ULONG, ZMOD_RW, "Maximum size in bytes allowed for src nvlist passed with ZFS ioctls"); ZFS_MODULE_PARAM(zfs, zfs_, history_output_max, ULONG, ZMOD_RW, "Maximum size in bytes of ZFS ioctl output that will be logged"); diff --git a/module/zfs/zio.c b/module/zfs/zio.c index cc2b61f2520b..c2e3c6169fa3 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,5073 +1,5073 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or https://opensource.org/licenses/CDDL-1.0. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2020 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2019, Klara Inc. * Copyright (c) 2019, Allan Jude * Copyright (c) 2021, Datto, Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *const zio_type_name[ZIO_TYPES] = { /* * Note: Linux kernel thread name length is limited * so these names will differ from upstream open zfs. */ "z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl", "z_trim" }; int zio_dva_throttle_enabled = B_TRUE; static int zio_deadman_log_all = B_FALSE; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ static kmem_cache_t *zio_cache; static kmem_cache_t *zio_link_cache; kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #if defined(ZFS_DEBUG) && !defined(_KERNEL) static uint64_t zio_buf_cache_allocs[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; static uint64_t zio_buf_cache_frees[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; #endif /* Mark IOs as "slow" if they take longer than 30 seconds */ static uint_t zio_slow_io_ms = (30 * MILLISEC); #define BP_SPANB(indblkshift, level) \ (((uint64_t)1) << ((level) * ((indblkshift) - SPA_BLKPTRSHIFT))) #define COMPARE_META_LEVEL 0x80000000ul /* * The following actions directly effect the spa's sync-to-convergence logic. * The values below define the sync pass when we start performing the action. * Care should be taken when changing these values as they directly impact * spa_sync() performance. Tuning these values may introduce subtle performance * pathologies and should only be done in the context of performance analysis. * These tunables will eventually be removed and replaced with #defines once * enough analysis has been done to determine optimal values. * * The 'zfs_sync_pass_deferred_free' pass must be greater than 1 to ensure that * regular blocks are not deferred. * * Starting in sync pass 8 (zfs_sync_pass_dont_compress), we disable * compression (including of metadata). In practice, we don't have this * many sync passes, so this has no effect. * * The original intent was that disabling compression would help the sync * passes to converge. However, in practice disabling compression increases * the average number of sync passes, because when we turn compression off, a * lot of block's size will change and thus we have to re-allocate (not * overwrite) them. It also increases the number of 128KB allocations (e.g. * for indirect blocks and spacemaps) because these will not be compressed. * The 128K allocations are especially detrimental to performance on highly * fragmented systems, which may have very few free segments of this size, * and may need to load new metaslabs to satisfy 128K allocations. */ /* defer frees starting in this pass */ uint_t zfs_sync_pass_deferred_free = 2; /* don't compress starting in this pass */ static uint_t zfs_sync_pass_dont_compress = 8; /* rewrite new bps starting in this pass */ static uint_t zfs_sync_pass_rewrite = 2; /* * An allocating zio is one that either currently has the DVA allocate * stage set or will have it later in its lifetime. */ #define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE) /* * Enable smaller cores by excluding metadata * allocations as well. */ int zio_exclude_metadata = 0; static int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG static const int zio_buf_debug_limit = 16384; #else static const int zio_buf_debug_limit = 0; #endif static inline void __zio_execute(zio_t *zio); static void zio_taskq_dispatch(zio_t *, zio_taskq_type_t, boolean_t); void zio_init(void) { size_t c; zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zio_link_cache = kmem_cache_create("zio_link_cache", sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0); /* * For small buffers, we want a cache for each multiple of * SPA_MINBLOCKSIZE. For larger buffers, we want a cache * for each quarter-power of 2. */ for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t p2 = size; size_t align = 0; size_t data_cflags, cflags; data_cflags = KMC_NODEBUG; cflags = (zio_exclude_metadata || size > zio_buf_debug_limit) ? KMC_NODEBUG : 0; while (!ISP2(p2)) p2 &= p2 - 1; #ifndef _KERNEL /* * If we are using watchpoints, put each buffer on its own page, * to eliminate the performance overhead of trapping to the * kernel when modifying a non-watched buffer that shares the * page with a watched buffer. */ if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE)) continue; /* * Here's the problem - on 4K native devices in userland on * Linux using O_DIRECT, buffers must be 4K aligned or I/O * will fail with EINVAL, causing zdb (and others) to coredump. * Since userland probably doesn't need optimized buffer caches, * we just force 4K alignment on everything. */ align = 8 * SPA_MINBLOCKSIZE; #else if (size < PAGESIZE) { align = SPA_MINBLOCKSIZE; } else if (IS_P2ALIGNED(size, p2 >> 2)) { align = PAGESIZE; } #endif if (align != 0) { char name[36]; if (cflags == data_cflags) { /* * Resulting kmem caches would be identical. * Save memory by creating only one. */ (void) snprintf(name, sizeof (name), "zio_buf_comb_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, cflags); zio_data_buf_cache[c] = zio_buf_cache[c]; continue; } (void) snprintf(name, sizeof (name), "zio_buf_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, cflags); (void) snprintf(name, sizeof (name), "zio_data_buf_%lu", (ulong_t)size); zio_data_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, data_cflags); } } while (--c != 0) { ASSERT(zio_buf_cache[c] != NULL); if (zio_buf_cache[c - 1] == NULL) zio_buf_cache[c - 1] = zio_buf_cache[c]; ASSERT(zio_data_buf_cache[c] != NULL); if (zio_data_buf_cache[c - 1] == NULL) zio_data_buf_cache[c - 1] = zio_data_buf_cache[c]; } zio_inject_init(); lz4_init(); } void zio_fini(void) { size_t n = SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; #if defined(ZFS_DEBUG) && !defined(_KERNEL) for (size_t i = 0; i < n; i++) { if (zio_buf_cache_allocs[i] != zio_buf_cache_frees[i]) (void) printf("zio_fini: [%d] %llu != %llu\n", (int)((i + 1) << SPA_MINBLOCKSHIFT), (long long unsigned)zio_buf_cache_allocs[i], (long long unsigned)zio_buf_cache_frees[i]); } #endif /* * The same kmem cache can show up multiple times in both zio_buf_cache * and zio_data_buf_cache. Do a wasteful but trivially correct scan to * sort it out. */ for (size_t i = 0; i < n; i++) { kmem_cache_t *cache = zio_buf_cache[i]; if (cache == NULL) continue; for (size_t j = i; j < n; j++) { if (cache == zio_buf_cache[j]) zio_buf_cache[j] = NULL; if (cache == zio_data_buf_cache[j]) zio_data_buf_cache[j] = NULL; } kmem_cache_destroy(cache); } for (size_t i = 0; i < n; i++) { kmem_cache_t *cache = zio_data_buf_cache[i]; if (cache == NULL) continue; for (size_t j = i; j < n; j++) { if (cache == zio_data_buf_cache[j]) zio_data_buf_cache[j] = NULL; } kmem_cache_destroy(cache); } for (size_t i = 0; i < n; i++) { VERIFY3P(zio_buf_cache[i], ==, NULL); VERIFY3P(zio_data_buf_cache[i], ==, NULL); } kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ /* * Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a * crashdump if the kernel panics, so use it judiciously. Obviously, it's * useful to inspect ZFS metadata, but if possible, we should avoid keeping * excess / transient data in-core during a crashdump. */ void * zio_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_allocs[c], 1); #endif return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE)); } /* * Use zio_data_buf_alloc to allocate data. The data will not appear in a * crashdump if the kernel panics. This exists so that we will limit the amount * of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount * of kernel heap dumped to disk when the kernel panics) */ void * zio_data_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE)); } void zio_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); #if defined(ZFS_DEBUG) && !defined(_KERNEL) atomic_add_64(&zio_buf_cache_frees[c], 1); #endif kmem_cache_free(zio_buf_cache[c], buf); } void zio_data_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_data_buf_cache[c], buf); } static void zio_abd_free(void *abd, size_t size) { (void) size; abd_free((abd_t *)abd); } /* * ========================================================================== * Push and pop I/O transform buffers * ========================================================================== */ void zio_push_transform(zio_t *zio, abd_t *data, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform) { zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_SLEEP); zt->zt_orig_abd = zio->io_abd; zt->zt_orig_size = zio->io_size; zt->zt_bufsize = bufsize; zt->zt_transform = transform; zt->zt_next = zio->io_transform_stack; zio->io_transform_stack = zt; zio->io_abd = data; zio->io_size = size; } void zio_pop_transforms(zio_t *zio) { zio_transform_t *zt; while ((zt = zio->io_transform_stack) != NULL) { if (zt->zt_transform != NULL) zt->zt_transform(zio, zt->zt_orig_abd, zt->zt_orig_size); if (zt->zt_bufsize != 0) abd_free(zio->io_abd); zio->io_abd = zt->zt_orig_abd; zio->io_size = zt->zt_orig_size; zio->io_transform_stack = zt->zt_next; kmem_free(zt, sizeof (zio_transform_t)); } } /* * ========================================================================== * I/O transform callbacks for subblocks, decompression, and decryption * ========================================================================== */ static void zio_subblock(zio_t *zio, abd_t *data, uint64_t size) { ASSERT(zio->io_size > size); if (zio->io_type == ZIO_TYPE_READ) abd_copy(data, zio->io_abd, size); } static void zio_decompress(zio_t *zio, abd_t *data, uint64_t size) { if (zio->io_error == 0) { void *tmp = abd_borrow_buf(data, size); int ret = zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_abd, tmp, zio->io_size, size, &zio->io_prop.zp_complevel); abd_return_buf_copy(data, tmp, size); if (zio_injection_enabled && ret == 0) ret = zio_handle_fault_injection(zio, EINVAL); if (ret != 0) zio->io_error = SET_ERROR(EIO); } } static void zio_decrypt(zio_t *zio, abd_t *data, uint64_t size) { int ret; void *tmp; blkptr_t *bp = zio->io_bp; spa_t *spa = zio->io_spa; uint64_t dsobj = zio->io_bookmark.zb_objset; uint64_t lsize = BP_GET_LSIZE(bp); dmu_object_type_t ot = BP_GET_TYPE(bp); uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; boolean_t no_crypt = B_FALSE; ASSERT(BP_USES_CRYPT(bp)); ASSERT3U(size, !=, 0); if (zio->io_error != 0) return; /* * Verify the cksum of MACs stored in an indirect bp. It will always * be possible to verify this since it does not require an encryption * key. */ if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) { zio_crypt_decode_mac_bp(bp, mac); if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF) { /* * We haven't decompressed the data yet, but * zio_crypt_do_indirect_mac_checksum() requires * decompressed data to be able to parse out the MACs * from the indirect block. We decompress it now and * throw away the result after we are finished. */ tmp = zio_buf_alloc(lsize); ret = zio_decompress_data(BP_GET_COMPRESS(bp), zio->io_abd, tmp, zio->io_size, lsize, &zio->io_prop.zp_complevel); if (ret != 0) { ret = SET_ERROR(EIO); goto error; } ret = zio_crypt_do_indirect_mac_checksum(B_FALSE, tmp, lsize, BP_SHOULD_BYTESWAP(bp), mac); zio_buf_free(tmp, lsize); } else { ret = zio_crypt_do_indirect_mac_checksum_abd(B_FALSE, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp), mac); } abd_copy(data, zio->io_abd, size); if (zio_injection_enabled && ot != DMU_OT_DNODE && ret == 0) { ret = zio_handle_decrypt_injection(spa, &zio->io_bookmark, ot, ECKSUM); } if (ret != 0) goto error; return; } /* * If this is an authenticated block, just check the MAC. It would be * nice to separate this out into its own flag, but for the moment * enum zio_flag is out of bits. */ if (BP_IS_AUTHENTICATED(bp)) { if (ot == DMU_OT_OBJSET) { ret = spa_do_crypt_objset_mac_abd(B_FALSE, spa, dsobj, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp)); } else { zio_crypt_decode_mac_bp(bp, mac); ret = spa_do_crypt_mac_abd(B_FALSE, spa, dsobj, zio->io_abd, size, mac); if (zio_injection_enabled && ret == 0) { ret = zio_handle_decrypt_injection(spa, &zio->io_bookmark, ot, ECKSUM); } } abd_copy(data, zio->io_abd, size); if (ret != 0) goto error; return; } zio_crypt_decode_params_bp(bp, salt, iv); if (ot == DMU_OT_INTENT_LOG) { tmp = abd_borrow_buf_copy(zio->io_abd, sizeof (zil_chain_t)); zio_crypt_decode_mac_zil(tmp, mac); abd_return_buf(zio->io_abd, tmp, sizeof (zil_chain_t)); } else { zio_crypt_decode_mac_bp(bp, mac); } ret = spa_do_crypt_abd(B_FALSE, spa, &zio->io_bookmark, BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp), salt, iv, mac, size, data, zio->io_abd, &no_crypt); if (no_crypt) abd_copy(data, zio->io_abd, size); if (ret != 0) goto error; return; error: /* assert that the key was found unless this was speculative */ ASSERT(ret != EACCES || (zio->io_flags & ZIO_FLAG_SPECULATIVE)); /* * If there was a decryption / authentication error return EIO as * the io_error. If this was not a speculative zio, create an ereport. */ if (ret == ECKSUM) { zio->io_error = SET_ERROR(EIO); if ((zio->io_flags & ZIO_FLAG_SPECULATIVE) == 0) { spa_log_error(spa, &zio->io_bookmark); (void) zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION, spa, NULL, &zio->io_bookmark, zio, 0); } } else { zio->io_error = ret; } } /* * ========================================================================== * I/O parent/child relationships and pipeline interlocks * ========================================================================== */ zio_t * zio_walk_parents(zio_t *cio, zio_link_t **zl) { list_t *pl = &cio->io_parent_list; *zl = (*zl == NULL) ? list_head(pl) : list_next(pl, *zl); if (*zl == NULL) return (NULL); ASSERT((*zl)->zl_child == cio); return ((*zl)->zl_parent); } zio_t * zio_walk_children(zio_t *pio, zio_link_t **zl) { list_t *cl = &pio->io_child_list; ASSERT(MUTEX_HELD(&pio->io_lock)); *zl = (*zl == NULL) ? list_head(cl) : list_next(cl, *zl); if (*zl == NULL) return (NULL); ASSERT((*zl)->zl_parent == pio); return ((*zl)->zl_child); } zio_t * zio_unique_parent(zio_t *cio) { zio_link_t *zl = NULL; zio_t *pio = zio_walk_parents(cio, &zl); VERIFY3P(zio_walk_parents(cio, &zl), ==, NULL); return (pio); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_SLEEP); /* * Logical I/Os can have logical, gang, or vdev children. * Gang I/Os can have gang or vdev children. * Vdev I/Os can only have vdev children. * The following ASSERT captures all of these constraints. */ ASSERT3S(cio->io_child_type, <=, pio->io_child_type); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&pio->io_lock); mutex_enter(&cio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); pio->io_child_count++; cio->io_parent_count++; mutex_exit(&cio->io_lock); mutex_exit(&pio->io_lock); } static void zio_remove_child(zio_t *pio, zio_t *cio, zio_link_t *zl) { ASSERT(zl->zl_parent == pio); ASSERT(zl->zl_child == cio); mutex_enter(&pio->io_lock); mutex_enter(&cio->io_lock); list_remove(&pio->io_child_list, zl); list_remove(&cio->io_parent_list, zl); pio->io_child_count--; cio->io_parent_count--; mutex_exit(&cio->io_lock); mutex_exit(&pio->io_lock); kmem_cache_free(zio_link_cache, zl); } static boolean_t zio_wait_for_children(zio_t *zio, uint8_t childbits, enum zio_wait_type wait) { boolean_t waiting = B_FALSE; mutex_enter(&zio->io_lock); ASSERT(zio->io_stall == NULL); for (int c = 0; c < ZIO_CHILD_TYPES; c++) { if (!(ZIO_CHILD_BIT_IS_SET(childbits, c))) continue; uint64_t *countp = &zio->io_children[c][wait]; if (*countp != 0) { zio->io_stage >>= 1; ASSERT3U(zio->io_stage, !=, ZIO_STAGE_OPEN); zio->io_stall = countp; waiting = B_TRUE; break; } } mutex_exit(&zio->io_lock); return (waiting); } __attribute__((always_inline)) static inline void zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait, zio_t **next_to_executep) { uint64_t *countp = &pio->io_children[zio->io_child_type][wait]; int *errorp = &pio->io_child_error[zio->io_child_type]; mutex_enter(&pio->io_lock); if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) *errorp = zio_worst_error(*errorp, zio->io_error); pio->io_reexecute |= zio->io_reexecute; ASSERT3U(*countp, >, 0); (*countp)--; if (*countp == 0 && pio->io_stall == countp) { zio_taskq_type_t type = pio->io_stage < ZIO_STAGE_VDEV_IO_START ? ZIO_TASKQ_ISSUE : ZIO_TASKQ_INTERRUPT; pio->io_stall = NULL; mutex_exit(&pio->io_lock); /* * If we can tell the caller to execute this parent next, do * so. Otherwise dispatch the parent zio as its own task. * * Having the caller execute the parent when possible reduces * locking on the zio taskq's, reduces context switch * overhead, and has no recursion penalty. Note that one * read from disk typically causes at least 3 zio's: a * zio_null(), the logical zio_read(), and then a physical * zio. When the physical ZIO completes, we are able to call * zio_done() on all 3 of these zio's from one invocation of * zio_execute() by returning the parent back to * zio_execute(). Since the parent isn't executed until this * thread returns back to zio_execute(), the caller should do * so promptly. * * In other cases, dispatching the parent prevents * overflowing the stack when we have deeply nested * parent-child relationships, as we do with the "mega zio" * of writes for spa_sync(), and the chain of ZIL blocks. */ if (next_to_executep != NULL && *next_to_executep == NULL) { *next_to_executep = pio; } else { zio_taskq_dispatch(pio, type, B_FALSE); } } else { mutex_exit(&pio->io_lock); } } static void zio_inherit_child_errors(zio_t *zio, enum zio_child c) { if (zio->io_child_error[c] != 0 && zio->io_error == 0) zio->io_error = zio->io_child_error[c]; } int zio_bookmark_compare(const void *x1, const void *x2) { const zio_t *z1 = x1; const zio_t *z2 = x2; if (z1->io_bookmark.zb_objset < z2->io_bookmark.zb_objset) return (-1); if (z1->io_bookmark.zb_objset > z2->io_bookmark.zb_objset) return (1); if (z1->io_bookmark.zb_object < z2->io_bookmark.zb_object) return (-1); if (z1->io_bookmark.zb_object > z2->io_bookmark.zb_object) return (1); if (z1->io_bookmark.zb_level < z2->io_bookmark.zb_level) return (-1); if (z1->io_bookmark.zb_level > z2->io_bookmark.zb_level) return (1); if (z1->io_bookmark.zb_blkid < z2->io_bookmark.zb_blkid) return (-1); if (z1->io_bookmark.zb_blkid > z2->io_bookmark.zb_blkid) return (1); if (z1 < z2) return (-1); if (z1 > z2) return (1); return (0); } /* * ========================================================================== * Create the various types of I/O (read, write, free, etc) * ========================================================================== */ static zio_t * zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, zio_done_func_t *done, void *private, zio_type_t type, zio_priority_t priority, enum zio_flag flags, vdev_t *vd, uint64_t offset, const zbookmark_phys_t *zb, enum zio_stage stage, enum zio_stage pipeline) { zio_t *zio; IMPLY(type != ZIO_TYPE_TRIM, psize <= SPA_MAXBLOCKSIZE); ASSERT(P2PHASE(psize, SPA_MINBLOCKSIZE) == 0); ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0); ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER)); ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER)); ASSERT(vd || stage == ZIO_STAGE_OPEN); IMPLY(lsize != psize, (flags & ZIO_FLAG_RAW_COMPRESS) != 0); zio = kmem_cache_alloc(zio_cache, KM_SLEEP); memset(zio, 0, sizeof (zio_t)); mutex_init(&zio->io_lock, NULL, MUTEX_NOLOCKDEP, NULL); cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL); list_create(&zio->io_parent_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_parent_node)); list_create(&zio->io_child_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_child_node)); metaslab_trace_init(&zio->io_alloc_list); if (vd != NULL) zio->io_child_type = ZIO_CHILD_VDEV; else if (flags & ZIO_FLAG_GANG_CHILD) zio->io_child_type = ZIO_CHILD_GANG; else if (flags & ZIO_FLAG_DDT_CHILD) zio->io_child_type = ZIO_CHILD_DDT; else zio->io_child_type = ZIO_CHILD_LOGICAL; if (bp != NULL) { zio->io_bp = (blkptr_t *)bp; zio->io_bp_copy = *bp; zio->io_bp_orig = *bp; if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) zio->io_bp = &zio->io_bp_copy; /* so caller can free */ if (zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_logical = zio; if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp)) pipeline |= ZIO_GANG_STAGES; } zio->io_spa = spa; zio->io_txg = txg; zio->io_done = done; zio->io_private = private; zio->io_type = type; zio->io_priority = priority; zio->io_vd = vd; zio->io_offset = offset; zio->io_orig_abd = zio->io_abd = data; zio->io_orig_size = zio->io_size = psize; zio->io_lsize = lsize; zio->io_orig_flags = zio->io_flags = flags; zio->io_orig_stage = zio->io_stage = stage; zio->io_orig_pipeline = zio->io_pipeline = pipeline; zio->io_pipeline_trace = ZIO_STAGE_OPEN; zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY); zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != NULL) { zio->io_metaslab_class = pio->io_metaslab_class; if (zio->io_logical == NULL) zio->io_logical = pio->io_logical; if (zio->io_child_type == ZIO_CHILD_GANG) zio->io_gang_leader = pio->io_gang_leader; zio_add_child(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } void zio_destroy(zio_t *zio) { metaslab_trace_fini(&zio->io_alloc_list); list_destroy(&zio->io_parent_list); list_destroy(&zio->io_child_list); mutex_destroy(&zio->io_lock); cv_destroy(&zio->io_cv); kmem_cache_free(zio_cache, zio); } zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE); return (zio); } zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags) { return (zio_null(NULL, spa, NULL, done, private, flags)); } static int zfs_blkptr_verify_log(spa_t *spa, const blkptr_t *bp, enum blk_verify_flag blk_verify, const char *fmt, ...) { va_list adx; char buf[256]; va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); switch (blk_verify) { case BLK_VERIFY_HALT: dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp); zfs_panic_recover("%s: %s", spa_name(spa), buf); break; case BLK_VERIFY_LOG: zfs_dbgmsg("%s: %s", spa_name(spa), buf); break; case BLK_VERIFY_ONLY: break; } return (1); } /* * Verify the block pointer fields contain reasonable values. This means * it only contains known object types, checksum/compression identifiers, * block sizes within the maximum allowed limits, valid DVAs, etc. * * If everything checks out B_TRUE is returned. The zfs_blkptr_verify * argument controls the behavior when an invalid field is detected. * * Modes for zfs_blkptr_verify: * 1) BLK_VERIFY_ONLY (evaluate the block) * 2) BLK_VERIFY_LOG (evaluate the block and log problems) * 3) BLK_VERIFY_HALT (call zfs_panic_recover on error) */ boolean_t zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp, boolean_t config_held, enum blk_verify_flag blk_verify) { int errors = 0; if (!DMU_OT_IS_VALID(BP_GET_TYPE(bp))) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid TYPE %llu", bp, (longlong_t)BP_GET_TYPE(bp)); } if (BP_GET_CHECKSUM(bp) >= ZIO_CHECKSUM_FUNCTIONS) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid CHECKSUM %llu", bp, (longlong_t)BP_GET_CHECKSUM(bp)); } if (BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_FUNCTIONS) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid COMPRESS %llu", bp, (longlong_t)BP_GET_COMPRESS(bp)); } if (BP_GET_LSIZE(bp) > SPA_MAXBLOCKSIZE) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid LSIZE %llu", bp, (longlong_t)BP_GET_LSIZE(bp)); } if (BP_GET_PSIZE(bp) > SPA_MAXBLOCKSIZE) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid PSIZE %llu", bp, (longlong_t)BP_GET_PSIZE(bp)); } if (BP_IS_EMBEDDED(bp)) { if (BPE_GET_ETYPE(bp) >= NUM_BP_EMBEDDED_TYPES) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p has invalid ETYPE %llu", bp, (longlong_t)BPE_GET_ETYPE(bp)); } } /* * Do not verify individual DVAs if the config is not trusted. This * will be done once the zio is executed in vdev_mirror_map_alloc. */ if (!spa->spa_trust_config) return (errors == 0); if (!config_held) spa_config_enter(spa, SCL_VDEV, bp, RW_READER); else ASSERT(spa_config_held(spa, SCL_VDEV, RW_WRITER)); /* * Pool-specific checks. * * Note: it would be nice to verify that the blk_birth and * BP_PHYSICAL_BIRTH() are not too large. However, spa_freeze() * allows the birth time of log blocks (and dmu_sync()-ed blocks * that are in the log) to be arbitrarily large. */ for (int i = 0; i < BP_GET_NDVAS(bp); i++) { const dva_t *dva = &bp->blk_dva[i]; uint64_t vdevid = DVA_GET_VDEV(dva); if (vdevid >= spa->spa_root_vdev->vdev_children) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_hole_ops) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has hole VDEV %llu", bp, i, (longlong_t)vdevid); continue; } if (vd->vdev_ops == &vdev_missing_ops) { /* * "missing" vdevs are valid during import, but we * don't have their detailed info (e.g. asize), so * we can't perform any more checks on them. */ continue; } uint64_t offset = DVA_GET_OFFSET(dva); uint64_t asize = DVA_GET_ASIZE(dva); if (DVA_GET_GANG(dva)) asize = vdev_gang_header_asize(vd); if (offset + asize > vd->vdev_asize) { errors += zfs_blkptr_verify_log(spa, bp, blk_verify, "blkptr at %p DVA %u has invalid OFFSET %llu", bp, i, (longlong_t)offset); } } if (errors > 0) dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp); if (!config_held) spa_config_exit(spa, SCL_VDEV, bp); return (errors == 0); } boolean_t zfs_dva_valid(spa_t *spa, const dva_t *dva, const blkptr_t *bp) { (void) bp; uint64_t vdevid = DVA_GET_VDEV(dva); if (vdevid >= spa->spa_root_vdev->vdev_children) return (B_FALSE); vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) return (B_FALSE); if (vd->vdev_ops == &vdev_hole_ops) return (B_FALSE); if (vd->vdev_ops == &vdev_missing_ops) { return (B_FALSE); } uint64_t offset = DVA_GET_OFFSET(dva); uint64_t asize = DVA_GET_ASIZE(dva); if (DVA_GET_GANG(dva)) asize = vdev_gang_header_asize(vd); if (offset + asize > vd->vdev_asize) return (B_FALSE); return (B_TRUE); } zio_t * zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp, data, size, size, done, private, ZIO_TYPE_READ, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE); return (zio); } zio_t * zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t lsize, uint64_t psize, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *children_ready, zio_done_func_t *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb) { zio_t *zio; ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF && zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS && zp->zp_compress >= ZIO_COMPRESS_OFF && zp->zp_compress < ZIO_COMPRESS_FUNCTIONS && DMU_OT_IS_VALID(zp->zp_type) && zp->zp_level < 32 && zp->zp_copies > 0 && zp->zp_copies <= spa_max_replication(spa)); zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE); zio->io_ready = ready; zio->io_children_ready = children_ready; zio->io_physdone = physdone; zio->io_prop = *zp; /* * Data can be NULL if we are going to call zio_write_override() to * provide the already-allocated BP. But we may need the data to * verify a dedup hit (if requested). In this case, don't try to * dedup (just take the already-allocated BP verbatim). Encrypted * dedup blocks need data as well so we also disable dedup in this * case. */ if (data == NULL && (zio->io_prop.zp_dedup_verify || zio->io_prop.zp_encrypt)) { zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE; } return (zio); } zio_t * zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, abd_t *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb) { zio_t *zio; zio = zio_create(pio, spa, txg, bp, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_IO_REWRITE, NULL, 0, zb, ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE); return (zio); } void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa)); /* * We must reset the io_prop to match the values that existed * when the bp was first written by dmu_sync() keeping in mind * that nopwrite and dedup are mutually exclusive. */ zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup; zio->io_prop.zp_nopwrite = nopwrite; zio->io_prop.zp_copies = copies; zio->io_bp_override = bp; } void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp) { (void) zfs_blkptr_verify(spa, bp, B_FALSE, BLK_VERIFY_HALT); /* * The check for EMBEDDED is a performance optimization. We * process the free here (by ignoring it) rather than * putting it on the list and then processing it in zio_free_sync(). */ if (BP_IS_EMBEDDED(bp)) return; metaslab_check_free(spa, bp); /* * Frees that are for the currently-syncing txg, are not going to be * deferred, and which will not need to do a read (i.e. not GANG or * DEDUP), can be processed immediately. Otherwise, put them on the * in-memory list for later processing. * * Note that we only defer frees after zfs_sync_pass_deferred_free * when the log space map feature is disabled. [see relevant comment * in spa_sync_iterate_to_convergence()] */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) || txg != spa->spa_syncing_txg || (spa_sync_pass(spa) >= zfs_sync_pass_deferred_free && !spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))) { bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp); } else { VERIFY3P(zio_free_sync(NULL, spa, txg, bp, 0), ==, NULL); } } /* * To improve performance, this function may return NULL if we were able * to do the free immediately. This avoids the cost of creating a zio * (and linking it to the parent, etc). */ zio_t * zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag flags) { ASSERT(!BP_IS_HOLE(bp)); ASSERT(spa_syncing_txg(spa) == txg); if (BP_IS_EMBEDDED(bp)) return (NULL); metaslab_check_free(spa, bp); arc_freed(spa, bp); dsl_scan_freed(spa, bp); if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp)) { /* * GANG and DEDUP blocks can induce a read (for the gang block * header, or the DDT), so issue them asynchronously so that * this thread is not tied up. */ enum zio_stage stage = ZIO_FREE_PIPELINE | ZIO_STAGE_ISSUE_ASYNC; return (zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), BP_GET_PSIZE(bp), NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage)); } else { metaslab_free(spa, bp, txg, B_FALSE); return (NULL); } } zio_t * zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; (void) zfs_blkptr_verify(spa, bp, flags & ZIO_FLAG_CONFIG_WRITER, BLK_VERIFY_HALT); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); /* * A claim is an allocation of a specific block. Claims are needed * to support immediate writes in the intent log. The issue is that * immediate writes contain committed data, but in a txg that was * *not* committed. Upon opening the pool after an unclean shutdown, * the intent log claims all blocks that contain immediate write data * so that the SPA knows they're in use. * * All claims *must* be resolved in the first txg -- before the SPA * starts allocating blocks -- so that nothing is allocated twice. * If txg == 0 we just verify that the block is claimable. */ ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <, spa_min_claim_txg(spa)); ASSERT(txg == spa_min_claim_txg(spa) || txg == 0); ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(8) */ zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE); ASSERT0(zio->io_queued_timestamp); return (zio); } zio_t * zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; int c; if (vd->vdev_children == 0) { zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private, ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE); zio->io_cmd = cmd; } else { zio = zio_null(pio, spa, NULL, NULL, NULL, flags); for (c = 0; c < vd->vdev_children; c++) zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd, done, private, flags)); } return (zio); } zio_t * zio_trim(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, enum trim_flag trim_flags) { zio_t *zio; ASSERT0(vd->vdev_children); ASSERT0(P2PHASE(offset, 1ULL << vd->vdev_ashift)); ASSERT0(P2PHASE(size, 1ULL << vd->vdev_ashift)); ASSERT3U(size, !=, 0); zio = zio_create(pio, vd->vdev_spa, 0, NULL, NULL, size, size, done, private, ZIO_TYPE_TRIM, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_TRIM_PIPELINE); zio->io_trim_flags = trim_flags; return (zio); } zio_t * zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; return (zio); } zio_t * zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, abd_t *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done, private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; if (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_EMBEDDED) { /* * zec checksums are necessarily destructive -- they modify * the end of the write buffer to hold the verifier/checksum. * Therefore, we must make a local copy in case the data is * being written to multiple places in parallel. */ abd_t *wbuf = abd_alloc_sametype(data, size); abd_copy(wbuf, data, size); zio_push_transform(zio, wbuf, size, size, NULL); } return (zio); } /* * Create a child I/O to do some work for us. */ zio_t * zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE; zio_t *zio; /* * vdev child I/Os do not propagate their error to the parent. * Therefore, for correct operation the caller *must* check for * and handle the error in the child i/o's done callback. * The only exceptions are i/os that we don't care about * (OPTIONAL or REPAIR). */ ASSERT((flags & ZIO_FLAG_OPTIONAL) || (flags & ZIO_FLAG_IO_REPAIR) || done != NULL); if (type == ZIO_TYPE_READ && bp != NULL) { /* * If we have the bp, then the child should perform the * checksum and the parent need not. This pushes error * detection as close to the leaves as possible and * eliminates redundant checksums in the interior nodes. */ pipeline |= ZIO_STAGE_CHECKSUM_VERIFY; pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } if (vd->vdev_ops->vdev_op_leaf) { ASSERT0(vd->vdev_children); offset += VDEV_LABEL_START_SIZE; } flags |= ZIO_VDEV_CHILD_FLAGS(pio); /* * If we've decided to do a repair, the write is not speculative -- * even if the original read was. */ if (flags & ZIO_FLAG_IO_REPAIR) flags &= ~ZIO_FLAG_SPECULATIVE; /* * If we're creating a child I/O that is not associated with a * top-level vdev, then the child zio is not an allocating I/O. * If this is a retried I/O then we ignore it since we will * have already processed the original allocating I/O. */ if (flags & ZIO_FLAG_IO_ALLOCATING && (vd != vd->vdev_top || (flags & ZIO_FLAG_IO_RETRY))) { ASSERT(pio->io_metaslab_class != NULL); ASSERT(pio->io_metaslab_class->mc_alloc_throttle_enabled); ASSERT(type == ZIO_TYPE_WRITE); ASSERT(priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(!(flags & ZIO_FLAG_IO_REPAIR)); ASSERT(!(pio->io_flags & ZIO_FLAG_IO_REWRITE) || pio->io_child_type == ZIO_CHILD_GANG); flags &= ~ZIO_FLAG_IO_ALLOCATING; } zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, size, done, private, type, priority, flags, vd, offset, &pio->io_bookmark, ZIO_STAGE_VDEV_IO_START >> 1, pipeline); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); zio->io_physdone = pio->io_physdone; if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL) zio->io_logical->io_phys_children++; return (zio); } zio_t * zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, abd_t *data, uint64_t size, zio_type_t type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { zio_t *zio; ASSERT(vd->vdev_ops->vdev_op_leaf); zio = zio_create(NULL, vd->vdev_spa, 0, NULL, data, size, size, done, private, type, priority, flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED, vd, offset, NULL, ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE); return (zio); } void zio_flush(zio_t *zio, vdev_t *vd) { zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY)); } void zio_shrink(zio_t *zio, uint64_t size) { ASSERT3P(zio->io_executor, ==, NULL); ASSERT3U(zio->io_orig_size, ==, zio->io_size); ASSERT3U(size, <=, zio->io_size); /* * We don't shrink for raidz because of problems with the * reconstruction when reading back less than the block size. * Note, BP_IS_RAIDZ() assumes no compression. */ ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); if (!BP_IS_RAIDZ(zio->io_bp)) { /* we are not doing a raw write */ ASSERT3U(zio->io_size, ==, zio->io_lsize); zio->io_orig_size = zio->io_size = zio->io_lsize = size; } } /* * ========================================================================== * Prepare to read and write logical blocks * ========================================================================== */ static zio_t * zio_read_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; uint64_t psize = BP_IS_EMBEDDED(bp) ? BPE_GET_PSIZE(bp) : BP_GET_PSIZE(bp); ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF && zio->io_child_type == ZIO_CHILD_LOGICAL && !(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) { zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize), psize, psize, zio_decompress); } if (((BP_IS_PROTECTED(bp) && !(zio->io_flags & ZIO_FLAG_RAW_ENCRYPT)) || BP_HAS_INDIRECT_MAC_CKSUM(bp)) && zio->io_child_type == ZIO_CHILD_LOGICAL) { zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize), psize, psize, zio_decrypt); } if (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA) { int psize = BPE_GET_PSIZE(bp); void *data = abd_borrow_buf(zio->io_abd, psize); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; decode_embedded_bp_compressed(bp, data); abd_return_buf_copy(zio->io_abd, data, psize); } else { ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); } if (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_pipeline = ZIO_DDT_READ_PIPELINE; return (zio); } static zio_t * zio_write_bp_init(zio_t *zio) { if (!IO_IS_ALLOCATING(zio)) return (zio); ASSERT(zio->io_child_type != ZIO_CHILD_DDT); if (zio->io_bp_override) { blkptr_t *bp = zio->io_bp; zio_prop_t *zp = &zio->io_prop; ASSERT(bp->blk_birth != zio->io_txg); ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (BP_IS_EMBEDDED(bp)) return (zio); /* * If we've been overridden and nopwrite is set then * set the flag accordingly to indicate that a nopwrite * has already occurred. */ if (!BP_IS_HOLE(bp) && zp->zp_nopwrite) { ASSERT(!zp->zp_dedup); ASSERT3U(BP_GET_CHECKSUM(bp), ==, zp->zp_checksum); zio->io_flags |= ZIO_FLAG_NOPWRITE; return (zio); } ASSERT(!zp->zp_nopwrite); if (BP_IS_HOLE(bp) || !zp->zp_dedup) return (zio); ASSERT((zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP) || zp->zp_dedup_verify); if (BP_GET_CHECKSUM(bp) == zp->zp_checksum && !zp->zp_encrypt) { BP_SET_DEDUP(bp, 1); zio->io_pipeline |= ZIO_STAGE_DDT_WRITE; return (zio); } /* * We were unable to handle this as an override bp, treat * it as a regular write I/O. */ zio->io_bp_override = NULL; *bp = zio->io_bp_orig; zio->io_pipeline = zio->io_orig_pipeline; } return (zio); } static zio_t * zio_write_compress(zio_t *zio) { spa_t *spa = zio->io_spa; zio_prop_t *zp = &zio->io_prop; enum zio_compress compress = zp->zp_compress; blkptr_t *bp = zio->io_bp; uint64_t lsize = zio->io_lsize; uint64_t psize = zio->io_size; uint32_t pass = 1; /* * If our children haven't all reached the ready stage, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_LOGICAL_BIT | ZIO_CHILD_GANG_BIT, ZIO_WAIT_READY)) { return (NULL); } if (!IO_IS_ALLOCATING(zio)) return (zio); if (zio->io_children_ready != NULL) { /* * Now that all our children are ready, run the callback * associated with this zio in case it wants to modify the * data to be written. */ ASSERT3U(zp->zp_level, >, 0); zio->io_children_ready(zio); } ASSERT(zio->io_child_type != ZIO_CHILD_DDT); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) { /* * We're rewriting an existing block, which means we're * working on behalf of spa_sync(). For spa_sync() to * converge, it must eventually be the case that we don't * have to allocate new blocks. But compression changes * the blocksize, which forces a reallocate, and makes * convergence take longer. Therefore, after the first * few passes, stop compressing to ensure convergence. */ pass = spa_sync_pass(spa); ASSERT(zio->io_txg == spa_syncing_txg(spa)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!BP_GET_DEDUP(bp)); if (pass >= zfs_sync_pass_dont_compress) compress = ZIO_COMPRESS_OFF; /* Make sure someone doesn't change their mind on overwrites */ ASSERT(BP_IS_EMBEDDED(bp) || MIN(zp->zp_copies + BP_IS_GANG(bp), spa_max_replication(spa)) == BP_GET_NDVAS(bp)); } /* If it's a compressed write that is not raw, compress the buffer. */ if (compress != ZIO_COMPRESS_OFF && !(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) { void *cbuf = zio_buf_alloc(lsize); psize = zio_compress_data(compress, zio->io_abd, cbuf, lsize, zp->zp_complevel); if (psize == 0 || psize >= lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else if (!zp->zp_dedup && !zp->zp_encrypt && psize <= BPE_PAYLOAD_SIZE && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { encode_embedded_bp_compressed(bp, cbuf, compress, lsize, psize); BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA); BP_SET_TYPE(bp, zio->io_prop.zp_type); BP_SET_LEVEL(bp, zio->io_prop.zp_level); zio_buf_free(cbuf, lsize); bp->blk_birth = zio->io_txg; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (zio); } else { /* * Round compressed size up to the minimum allocation * size of the smallest-ashift device, and zero the * tail. This ensures that the compressed size of the * BP (and thus compressratio property) are correct, * in that we charge for the padding used to fill out * the last sector. */ ASSERT3U(spa->spa_min_alloc, >=, SPA_MINBLOCKSHIFT); size_t rounded = (size_t)roundup(psize, spa->spa_min_alloc); if (rounded >= lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); psize = lsize; } else { abd_t *cdata = abd_get_from_buf(cbuf, lsize); abd_take_ownership_of_buf(cdata, B_TRUE); abd_zero_off(cdata, psize, rounded - psize); psize = rounded; zio_push_transform(zio, cdata, psize, lsize, NULL); } } /* * We were unable to handle this as an override bp, treat * it as a regular write I/O. */ zio->io_bp_override = NULL; *bp = zio->io_bp_orig; zio->io_pipeline = zio->io_orig_pipeline; } else if ((zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) != 0 && zp->zp_type == DMU_OT_DNODE) { /* * The DMU actually relies on the zio layer's compression * to free metadnode blocks that have had all contained * dnodes freed. As a result, even when doing a raw * receive, we must check whether the block can be compressed * to a hole. */ psize = zio_compress_data(ZIO_COMPRESS_EMPTY, zio->io_abd, NULL, lsize, zp->zp_complevel); if (psize == 0 || psize >= lsize) compress = ZIO_COMPRESS_OFF; } else if (zio->io_flags & ZIO_FLAG_RAW_COMPRESS && !(zio->io_flags & ZIO_FLAG_RAW_ENCRYPT)) { /* * If we are raw receiving an encrypted dataset we should not * take this codepath because it will change the on-disk block * and decryption will fail. */ size_t rounded = MIN((size_t)roundup(psize, spa->spa_min_alloc), lsize); if (rounded != psize) { abd_t *cdata = abd_alloc_linear(rounded, B_TRUE); abd_zero_off(cdata, psize, rounded - psize); abd_copy_off(cdata, zio->io_abd, 0, 0, psize); psize = rounded; zio_push_transform(zio, cdata, psize, rounded, NULL); } } else { ASSERT3U(psize, !=, 0); } /* * The final pass of spa_sync() must be all rewrites, but the first * few passes offer a trade-off: allocating blocks defers convergence, * but newly allocated blocks are sequential, so they can be written * to disk faster. Therefore, we allow the first few passes of * spa_sync() to allocate new blocks, but force rewrites after that. * There should only be a handful of blocks after pass 1 in any case. */ if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg && BP_GET_PSIZE(bp) == psize && pass >= zfs_sync_pass_rewrite) { VERIFY3U(psize, !=, 0); enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES; zio->io_pipeline = ZIO_REWRITE_PIPELINE | gang_stages; zio->io_flags |= ZIO_FLAG_IO_REWRITE; } else { BP_ZERO(bp); zio->io_pipeline = ZIO_WRITE_PIPELINE; } if (psize == 0) { if (zio->io_bp_orig.blk_birth != 0 && spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_BIRTH(bp, zio->io_txg, 0); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } else { ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER); BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_PSIZE(bp, psize); BP_SET_COMPRESS(bp, compress); BP_SET_CHECKSUM(bp, zp->zp_checksum); BP_SET_DEDUP(bp, zp->zp_dedup); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); if (zp->zp_dedup) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!zp->zp_encrypt || DMU_OT_IS_ENCRYPTED(zp->zp_type)); zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE; } if (zp->zp_nopwrite) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline |= ZIO_STAGE_NOP_WRITE; } } return (zio); } static zio_t * zio_free_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio->io_child_type == ZIO_CHILD_LOGICAL) { if (BP_GET_DEDUP(bp)) zio->io_pipeline = ZIO_DDT_FREE_PIPELINE; } ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy); return (zio); } /* * ========================================================================== * Execute the I/O pipeline * ========================================================================== */ static void zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline) { spa_t *spa = zio->io_spa; zio_type_t t = zio->io_type; int flags = (cutinline ? TQ_FRONT : 0); /* * If we're a config writer or a probe, the normal issue and * interrupt threads may all be blocked waiting for the config lock. * In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL. */ if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_PROBE)) t = ZIO_TYPE_NULL; /* * A similar issue exists for the L2ARC write thread until L2ARC 2.0. */ if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux) t = ZIO_TYPE_NULL; /* * If this is a high priority I/O, then use the high priority taskq if * available. */ if ((zio->io_priority == ZIO_PRIORITY_NOW || zio->io_priority == ZIO_PRIORITY_SYNC_WRITE) && spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; ASSERT3U(q, <, ZIO_TASKQ_TYPES); /* * NB: We are assuming that the zio can only be dispatched * to a single taskq at a time. It would be a grievous error * to dispatch the zio to another taskq at the same time. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(spa, t, q, zio_execute, zio, flags, &zio->io_tqent); } static boolean_t zio_taskq_member(zio_t *zio, zio_taskq_type_t q) { spa_t *spa = zio->io_spa; taskq_t *tq = taskq_of_curthread(); for (zio_type_t t = 0; t < ZIO_TYPES; t++) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; for (i = 0; i < tqs->stqs_count; i++) { if (tqs->stqs_taskq[i] == tq) return (B_TRUE); } } return (B_FALSE); } static zio_t * zio_issue_async(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (NULL); } void zio_interrupt(void *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE); } void zio_delay_interrupt(zio_t *zio) { /* * The timeout_generic() function isn't defined in userspace, so * rather than trying to implement the function, the zio delay * functionality has been disabled for userspace builds. */ #ifdef _KERNEL /* * If io_target_timestamp is zero, then no delay has been registered * for this IO, thus jump to the end of this function and "skip" the * delay; issuing it directly to the zio layer. */ if (zio->io_target_timestamp != 0) { hrtime_t now = gethrtime(); if (now >= zio->io_target_timestamp) { /* * This IO has already taken longer than the target * delay to complete, so we don't want to delay it * any longer; we "miss" the delay and issue it * directly to the zio layer. This is likely due to * the target latency being set to a value less than * the underlying hardware can satisfy (e.g. delay * set to 1ms, but the disks take 10ms to complete an * IO request). */ DTRACE_PROBE2(zio__delay__miss, zio_t *, zio, hrtime_t, now); zio_interrupt(zio); } else { taskqid_t tid; hrtime_t diff = zio->io_target_timestamp - now; clock_t expire_at_tick = ddi_get_lbolt() + NSEC_TO_TICK(diff); DTRACE_PROBE3(zio__delay__hit, zio_t *, zio, hrtime_t, now, hrtime_t, diff); if (NSEC_TO_TICK(diff) == 0) { /* Our delay is less than a jiffy - just spin */ zfs_sleep_until(zio->io_target_timestamp); zio_interrupt(zio); } else { /* * Use taskq_dispatch_delay() in the place of * OpenZFS's timeout_generic(). */ tid = taskq_dispatch_delay(system_taskq, zio_interrupt, zio, TQ_NOSLEEP, expire_at_tick); if (tid == TASKQID_INVALID) { /* * Couldn't allocate a task. Just * finish the zio without a delay. */ zio_interrupt(zio); } } } return; } #endif DTRACE_PROBE1(zio__delay__skip, zio_t *, zio); zio_interrupt(zio); } static void zio_deadman_impl(zio_t *pio, int ziodepth) { zio_t *cio, *cio_next; zio_link_t *zl = NULL; vdev_t *vd = pio->io_vd; if (zio_deadman_log_all || (vd != NULL && vd->vdev_ops->vdev_op_leaf)) { vdev_queue_t *vq = vd ? &vd->vdev_queue : NULL; zbookmark_phys_t *zb = &pio->io_bookmark; uint64_t delta = gethrtime() - pio->io_timestamp; uint64_t failmode = spa_get_deadman_failmode(pio->io_spa); zfs_dbgmsg("slow zio[%d]: zio=%px timestamp=%llu " "delta=%llu queued=%llu io=%llu " "path=%s " "last=%llu type=%d " "priority=%d flags=0x%x stage=0x%x " "pipeline=0x%x pipeline-trace=0x%x " "objset=%llu object=%llu " "level=%llu blkid=%llu " "offset=%llu size=%llu " "error=%d", ziodepth, pio, pio->io_timestamp, (u_longlong_t)delta, pio->io_delta, pio->io_delay, vd ? vd->vdev_path : "NULL", vq ? vq->vq_io_complete_ts : 0, pio->io_type, pio->io_priority, pio->io_flags, pio->io_stage, pio->io_pipeline, pio->io_pipeline_trace, (u_longlong_t)zb->zb_objset, (u_longlong_t)zb->zb_object, (u_longlong_t)zb->zb_level, (u_longlong_t)zb->zb_blkid, (u_longlong_t)pio->io_offset, (u_longlong_t)pio->io_size, pio->io_error); (void) zfs_ereport_post(FM_EREPORT_ZFS_DEADMAN, pio->io_spa, vd, zb, pio, 0); if (failmode == ZIO_FAILURE_MODE_CONTINUE && taskq_empty_ent(&pio->io_tqent)) { zio_interrupt(pio); } } mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); zio_deadman_impl(cio, ziodepth + 1); } mutex_exit(&pio->io_lock); } /* * Log the critical information describing this zio and all of its children * using the zfs_dbgmsg() interface then post deadman event for the ZED. */ void zio_deadman(zio_t *pio, const char *tag) { spa_t *spa = pio->io_spa; char *name = spa_name(spa); if (!zfs_deadman_enabled || spa_suspended(spa)) return; zio_deadman_impl(pio, 0); switch (spa_get_deadman_failmode(spa)) { case ZIO_FAILURE_MODE_WAIT: zfs_dbgmsg("%s waiting for hung I/O to pool '%s'", tag, name); break; case ZIO_FAILURE_MODE_CONTINUE: zfs_dbgmsg("%s restarting hung I/O for pool '%s'", tag, name); break; case ZIO_FAILURE_MODE_PANIC: fm_panic("%s determined I/O to pool '%s' is hung.", tag, name); break; } } /* * Execute the I/O pipeline until one of the following occurs: * (1) the I/O completes; (2) the pipeline stalls waiting for * dependent child I/Os; (3) the I/O issues, so we're waiting * for an I/O completion interrupt; (4) the I/O is delegated by * vdev-level caching or aggregation; (5) the I/O is deferred * due to vdev-level queueing; (6) the I/O is handed off to * another thread. In all cases, the pipeline stops whenever * there's no CPU work; it never burns a thread in cv_wait_io(). * * There's no locking on io_stage because there's no legitimate way * for multiple threads to be attempting to process the same I/O. */ static zio_pipe_stage_t *zio_pipeline[]; /* * zio_execute() is a wrapper around the static function * __zio_execute() so that we can force __zio_execute() to be * inlined. This reduces stack overhead which is important * because __zio_execute() is called recursively in several zio * code paths. zio_execute() itself cannot be inlined because * it is externally visible. */ void zio_execute(void *zio) { fstrans_cookie_t cookie; cookie = spl_fstrans_mark(); __zio_execute(zio); spl_fstrans_unmark(cookie); } /* * Used to determine if in the current context the stack is sized large * enough to allow zio_execute() to be called recursively. A minimum * stack size of 16K is required to avoid needing to re-dispatch the zio. */ static boolean_t zio_execute_stack_check(zio_t *zio) { #if !defined(HAVE_LARGE_STACKS) dsl_pool_t *dp = spa_get_dsl(zio->io_spa); /* Executing in txg_sync_thread() context. */ if (dp && curthread == dp->dp_tx.tx_sync_thread) return (B_TRUE); /* Pool initialization outside of zio_taskq context. */ if (dp && spa_is_initializing(dp->dp_spa) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH)) return (B_TRUE); #else (void) zio; #endif /* HAVE_LARGE_STACKS */ return (B_FALSE); } __attribute__((always_inline)) static inline void __zio_execute(zio_t *zio) { ASSERT3U(zio->io_queued_timestamp, >, 0); while (zio->io_stage < ZIO_STAGE_DONE) { enum zio_stage pipeline = zio->io_pipeline; enum zio_stage stage = zio->io_stage; zio->io_executor = curthread; ASSERT(!MUTEX_HELD(&zio->io_lock)); ASSERT(ISP2(stage)); ASSERT(zio->io_stall == NULL); do { stage <<= 1; } while ((stage & pipeline) == 0); ASSERT(stage <= ZIO_STAGE_DONE); /* * If we are in interrupt context and this pipeline stage * will grab a config lock that is held across I/O, * or may wait for an I/O that needs an interrupt thread * to complete, issue async to avoid deadlock. * * For VDEV_IO_START, we cut in line so that the io will * be sent to disk promptly. */ if ((stage & ZIO_BLOCKING_STAGES) && zio->io_vd == NULL && zio_taskq_member(zio, ZIO_TASKQ_INTERRUPT)) { boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } /* * If the current context doesn't have large enough stacks * the zio must be issued asynchronously to prevent overflow. */ if (zio_execute_stack_check(zio)) { boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } zio->io_stage = stage; zio->io_pipeline_trace |= zio->io_stage; /* * The zio pipeline stage returns the next zio to execute * (typically the same as this one), or NULL if we should * stop. */ zio = zio_pipeline[highbit64(stage) - 1](zio); if (zio == NULL) return; } } /* * ========================================================================== * Initiate I/O, either sync or async * ========================================================================== */ int zio_wait(zio_t *zio) { /* * Some routines, like zio_free_sync(), may return a NULL zio * to avoid the performance overhead of creating and then destroying * an unneeded zio. For the callers' simplicity, we accept a NULL * zio and ignore it. */ if (zio == NULL) return (0); long timeout = MSEC_TO_TICK(zfs_deadman_ziotime_ms); int error; ASSERT3S(zio->io_stage, ==, ZIO_STAGE_OPEN); ASSERT3P(zio->io_executor, ==, NULL); zio->io_waiter = curthread; ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); mutex_enter(&zio->io_lock); while (zio->io_executor != NULL) { error = cv_timedwait_io(&zio->io_cv, &zio->io_lock, ddi_get_lbolt() + timeout); if (zfs_deadman_enabled && error == -1 && gethrtime() - zio->io_queued_timestamp > spa_deadman_ziotime(zio->io_spa)) { mutex_exit(&zio->io_lock); timeout = MSEC_TO_TICK(zfs_deadman_checktime_ms); zio_deadman(zio, FTAG); mutex_enter(&zio->io_lock); } } mutex_exit(&zio->io_lock); error = zio->io_error; zio_destroy(zio); return (error); } void zio_nowait(zio_t *zio) { /* * See comment in zio_wait(). */ if (zio == NULL) return; ASSERT3P(zio->io_executor, ==, NULL); if (zio->io_child_type == ZIO_CHILD_LOGICAL && zio_unique_parent(zio) == NULL) { zio_t *pio; /* * This is a logical async I/O with no parent to wait for it. * We add it to the spa_async_root_zio "Godfather" I/O which * will ensure they complete prior to unloading the pool. */ spa_t *spa = zio->io_spa; pio = spa->spa_async_zio_root[CPU_SEQID_UNSTABLE]; zio_add_child(pio, zio); } ASSERT0(zio->io_queued_timestamp); zio->io_queued_timestamp = gethrtime(); __zio_execute(zio); } /* * ========================================================================== * Reexecute, cancel, or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(void *arg) { zio_t *pio = arg; zio_t *cio, *cio_next; ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN); ASSERT(pio->io_gang_leader == NULL); ASSERT(pio->io_gang_tree == NULL); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_reexecute = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_pipeline_trace = 0; pio->io_error = 0; for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_state[w] = 0; for (int c = 0; c < ZIO_CHILD_TYPES; c++) pio->io_child_error[c] = 0; if (IO_IS_ALLOCATING(pio)) BP_ZERO(pio->io_bp); /* * As we reexecute pio's children, new children could be created. * New children go to the head of pio's io_child_list, however, * so we will (correctly) not reexecute them. The key is that * the remainder of pio's io_child_list, from 'cio_next' onward, * cannot be affected by any side effects of reexecuting 'cio'. */ zio_link_t *zl = NULL; mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); for (int w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w]++; mutex_exit(&pio->io_lock); zio_reexecute(cio); mutex_enter(&pio->io_lock); } mutex_exit(&pio->io_lock); /* * Now that all children have been reexecuted, execute the parent. * We don't reexecute "The Godfather" I/O here as it's the * responsibility of the caller to wait on it. */ if (!(pio->io_flags & ZIO_FLAG_GODFATHER)) { pio->io_queued_timestamp = gethrtime(); __zio_execute(pio); } } void zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t reason) { if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC) fm_panic("Pool '%s' has encountered an uncorrectable I/O " "failure and the failure mode property for this pool " "is set to panic.", spa_name(spa)); cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O " "failure and has been suspended.\n", spa_name(spa)); (void) zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL, NULL, NULL, 0); mutex_enter(&spa->spa_suspend_lock); if (spa->spa_suspend_zio_root == NULL) spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); spa->spa_suspended = reason; if (zio != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); ASSERT(zio != spa->spa_suspend_zio_root); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio_unique_parent(zio) == NULL); ASSERT(zio->io_stage == ZIO_STAGE_DONE); zio_add_child(spa->spa_suspend_zio_root, zio); } mutex_exit(&spa->spa_suspend_lock); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); spa->spa_suspended = ZIO_SUSPEND_NONE; cv_broadcast(&spa->spa_suspend_cv); pio = spa->spa_suspend_zio_root; spa->spa_suspend_zio_root = NULL; mutex_exit(&spa->spa_suspend_lock); if (pio == NULL) return (0); zio_reexecute(pio); return (zio_wait(pio)); } void zio_resume_wait(spa_t *spa) { mutex_enter(&spa->spa_suspend_lock); while (spa_suspended(spa)) cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock); mutex_exit(&spa->spa_suspend_lock); } /* * ========================================================================== * Gang blocks. * * A gang block is a collection of small blocks that looks to the DMU * like one large block. When zio_dva_allocate() cannot find a block * of the requested size, due to either severe fragmentation or the pool * being nearly full, it calls zio_write_gang_block() to construct the * block from smaller fragments. * * A gang block consists of a gang header (zio_gbh_phys_t) and up to * three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like * an indirect block: it's an array of block pointers. It consumes * only one sector and hence is allocatable regardless of fragmentation. * The gang header's bps point to its gang members, which hold the data. * * Gang blocks are self-checksumming, using the bp's * as the verifier to ensure uniqueness of the SHA256 checksum. * Critically, the gang block bp's blk_cksum is the checksum of the data, * not the gang header. This ensures that data block signatures (needed for * deduplication) are independent of how the block is physically stored. * * Gang blocks can be nested: a gang member may itself be a gang block. * Thus every gang block is a tree in which root and all interior nodes are * gang headers, and the leaves are normal blocks that contain user data. * The root of the gang tree is called the gang leader. * * To perform any operation (read, rewrite, free, claim) on a gang block, * zio_gang_assemble() first assembles the gang tree (minus data leaves) * in the io_gang_tree field of the original logical i/o by recursively * reading the gang leader and all gang headers below it. This yields * an in-core tree containing the contents of every gang header and the * bps for every constituent of the gang block. * * With the gang tree now assembled, zio_gang_issue() just walks the gang tree * and invokes a callback on each bp. To free a gang block, zio_gang_issue() * calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp. * zio_claim_gang() provides a similarly trivial wrapper for zio_claim(). * zio_read_gang() is a wrapper around zio_read() that omits reading gang * headers, since we already have those in io_gang_tree. zio_rewrite_gang() * performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite() * of the gang header plus zio_checksum_compute() of the data to update the * gang header's blk_cksum as described above. * * The two-phase assemble/issue model solves the problem of partial failure -- * what if you'd freed part of a gang block but then couldn't read the * gang header for another part? Assembling the entire gang tree first * ensures that all the necessary gang header I/O has succeeded before * starting the actual work of free, claim, or write. Once the gang tree * is assembled, free and claim are in-memory operations that cannot fail. * * In the event that a gang write fails, zio_dva_unallocate() walks the * gang tree to immediately free (i.e. insert back into the space map) * everything we've allocated. This ensures that we don't get ENOSPC * errors during repeated suspend/resume cycles due to a flaky device. * * Gang rewrites only happen during sync-to-convergence. If we can't assemble * the gang tree, we won't modify the block, so we can safely defer the free * (knowing that the block is still intact). If we *can* assemble the gang * tree, then even if some of the rewrites fail, zio_dva_unallocate() will free * each constituent bp and we can allocate a new block on the next sync pass. * * In all cases, the gang tree allows complete recovery from partial failure. * ========================================================================== */ static void zio_gang_issue_func_done(zio_t *zio) { abd_free(zio->io_abd); } static zio_t * zio_read_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { if (gn != NULL) return (pio); return (zio_read(pio, pio->io_spa, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } static zio_t * zio_rewrite_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { zio_t *zio; if (gn != NULL) { abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * As we rewrite each gang header, the pipeline will compute * a new gang block header checksum for it; but no one will * compute a new data checksum, so we do that here. The one * exception is the gang leader: the pipeline already computed * its data checksum because that stage precedes gang assembly. * (Presently, nothing actually uses interior data checksums; * this is just good hygiene.) */ if (gn != pio->io_gang_leader->io_gang_tree) { abd_t *buf = abd_get_offset(data, offset); zio_checksum_compute(zio, BP_GET_CHECKSUM(bp), buf, BP_GET_PSIZE(bp)); abd_free(buf); } /* * If we are here to damage data for testing purposes, * leave the GBH alone so that we can detect the damage. */ if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE) zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } else { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, abd_get_offset(data, offset), BP_GET_PSIZE(bp), zio_gang_issue_func_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); } return (zio); } static zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { (void) gn, (void) data, (void) offset; zio_t *zio = zio_free_sync(pio, pio->io_spa, pio->io_txg, bp, ZIO_GANG_CHILD_FLAGS(pio)); if (zio == NULL) { zio = zio_null(pio, pio->io_spa, NULL, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio)); } return (zio); } static zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, abd_t *data, uint64_t offset) { (void) gn, (void) data, (void) offset; return (zio_claim(pio, pio->io_spa, pio->io_txg, bp, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio))); } static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = { NULL, zio_read_gang, zio_rewrite_gang, zio_free_gang, zio_claim_gang, NULL }; static void zio_gang_tree_assemble_done(zio_t *zio); static zio_gang_node_t * zio_gang_node_alloc(zio_gang_node_t **gnpp) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn), KM_SLEEP); gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE); *gnpp = gn; return (gn); } static void zio_gang_node_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE); kmem_free(gn, sizeof (*gn)); *gnpp = NULL; } static void zio_gang_tree_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; if (gn == NULL) return; for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) zio_gang_tree_free(&gn->gn_child[g]); zio_gang_node_free(gnpp); } static void zio_gang_tree_assemble(zio_t *gio, blkptr_t *bp, zio_gang_node_t **gnpp) { zio_gang_node_t *gn = zio_gang_node_alloc(gnpp); abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_tree_assemble_done, gn, gio->io_priority, ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark)); } static void zio_gang_tree_assemble_done(zio_t *zio) { zio_t *gio = zio->io_gang_leader; zio_gang_node_t *gn = zio->io_private; blkptr_t *bp = zio->io_bp; ASSERT(gio == zio_unique_parent(zio)); ASSERT(zio->io_child_count == 0); if (zio->io_error) return; /* this ABD was created from a linear buf in zio_gang_tree_assemble */ if (BP_SHOULD_BYTESWAP(bp)) byteswap_uint64_array(abd_to_buf(zio->io_abd), zio->io_size); ASSERT3P(abd_to_buf(zio->io_abd), ==, gn->gn_gbh); ASSERT(zio->io_size == SPA_GANGBLOCKSIZE); ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); abd_free(zio->io_abd); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (!BP_IS_GANG(gbp)) continue; zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]); } } static void zio_gang_tree_issue(zio_t *pio, zio_gang_node_t *gn, blkptr_t *bp, abd_t *data, uint64_t offset) { zio_t *gio = pio->io_gang_leader; zio_t *zio; ASSERT(BP_IS_GANG(bp) == !!gn); ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp)); ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) || gn == gio->io_gang_tree); /* * If you're a gang header, your data is in gn->gn_gbh. * If you're a gang member, your data is in 'data' and gn == NULL. */ zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data, offset); if (gn != NULL) { ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (BP_IS_HOLE(gbp)) continue; zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data, offset); offset += BP_GET_PSIZE(gbp); } } if (gn == gio->io_gang_tree) ASSERT3U(gio->io_size, ==, offset); if (zio != pio) zio_nowait(zio); } static zio_t * zio_gang_assemble(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == NULL); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; zio_gang_tree_assemble(zio, bp, &zio->io_gang_tree); return (zio); } static zio_t * zio_gang_issue(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == zio); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); if (zio->io_child_error[ZIO_CHILD_GANG] == 0) zio_gang_tree_issue(zio, zio->io_gang_tree, bp, zio->io_abd, 0); else zio_gang_tree_free(&zio->io_gang_tree); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; return (zio); } static void zio_write_gang_member_ready(zio_t *zio) { zio_t *pio = zio_unique_parent(zio); dva_t *cdva = zio->io_bp->blk_dva; dva_t *pdva = pio->io_bp->blk_dva; uint64_t asize; zio_t *gio __maybe_unused = zio->io_gang_leader; if (BP_IS_HOLE(zio->io_bp)) return; ASSERT(BP_IS_HOLE(&zio->io_bp_orig)); ASSERT(zio->io_child_type == ZIO_CHILD_GANG); ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp)); ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp)); mutex_enter(&pio->io_lock); for (int d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) { ASSERT(DVA_GET_GANG(&pdva[d])); asize = DVA_GET_ASIZE(&pdva[d]); asize += DVA_GET_ASIZE(&cdva[d]); DVA_SET_ASIZE(&pdva[d], asize); } mutex_exit(&pio->io_lock); } static void zio_write_gang_done(zio_t *zio) { /* * The io_abd field will be NULL for a zio with no data. The io_flags * will initially have the ZIO_FLAG_NODATA bit flag set, but we can't * check for it here as it is cleared in zio_ready. */ if (zio->io_abd != NULL) abd_free(zio->io_abd); } static zio_t * zio_write_gang_block(zio_t *pio, metaslab_class_t *mc) { spa_t *spa = pio->io_spa; blkptr_t *bp = pio->io_bp; zio_t *gio = pio->io_gang_leader; zio_t *zio; zio_gang_node_t *gn, **gnpp; zio_gbh_phys_t *gbh; abd_t *gbh_abd; uint64_t txg = pio->io_txg; uint64_t resid = pio->io_size; uint64_t lsize; int copies = gio->io_prop.zp_copies; int gbh_copies; zio_prop_t zp; int error; boolean_t has_data = !(pio->io_flags & ZIO_FLAG_NODATA); /* * encrypted blocks need DVA[2] free so encrypted gang headers can't * have a third copy. */ gbh_copies = MIN(copies + 1, spa_max_replication(spa)); if (gio->io_prop.zp_encrypt && gbh_copies >= SPA_DVAS_PER_BP) gbh_copies = SPA_DVAS_PER_BP - 1; int flags = METASLAB_HINTBP_FAVOR | METASLAB_GANG_HEADER; if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); flags |= METASLAB_ASYNC_ALLOC; VERIFY(zfs_refcount_held(&mc->mc_allocator[pio->io_allocator]. mca_alloc_slots, pio)); /* * The logical zio has already placed a reservation for * 'copies' allocation slots but gang blocks may require * additional copies. These additional copies * (i.e. gbh_copies - copies) are guaranteed to succeed * since metaslab_class_throttle_reserve() always allows * additional reservations for gang blocks. */ VERIFY(metaslab_class_throttle_reserve(mc, gbh_copies - copies, pio->io_allocator, pio, flags)); } error = metaslab_alloc(spa, mc, SPA_GANGBLOCKSIZE, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, flags, &pio->io_alloc_list, pio, pio->io_allocator); if (error) { if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); /* * If we failed to allocate the gang block header then * we remove any additional allocation reservations that * we placed here. The original reservation will * be removed when the logical I/O goes to the ready * stage. */ metaslab_class_throttle_unreserve(mc, gbh_copies - copies, pio->io_allocator, pio); } pio->io_error = error; return (pio); } if (pio == gio) { gnpp = &gio->io_gang_tree; } else { gnpp = pio->io_private; ASSERT(pio->io_ready == zio_write_gang_member_ready); } gn = zio_gang_node_alloc(gnpp); gbh = gn->gn_gbh; memset(gbh, 0, SPA_GANGBLOCKSIZE); gbh_abd = abd_get_from_buf(gbh, SPA_GANGBLOCKSIZE); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh_abd, SPA_GANGBLOCKSIZE, zio_write_gang_done, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * Create and nowait the gang children. */ for (int g = 0; resid != 0; resid -= lsize, g++) { lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g), SPA_MINBLOCKSIZE); ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid); zp.zp_checksum = gio->io_prop.zp_checksum; zp.zp_compress = ZIO_COMPRESS_OFF; zp.zp_complevel = gio->io_prop.zp_complevel; zp.zp_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_dedup = B_FALSE; zp.zp_dedup_verify = B_FALSE; zp.zp_nopwrite = B_FALSE; zp.zp_encrypt = gio->io_prop.zp_encrypt; zp.zp_byteorder = gio->io_prop.zp_byteorder; memset(zp.zp_salt, 0, ZIO_DATA_SALT_LEN); memset(zp.zp_iv, 0, ZIO_DATA_IV_LEN); memset(zp.zp_mac, 0, ZIO_DATA_MAC_LEN); zio_t *cio = zio_write(zio, spa, txg, &gbh->zg_blkptr[g], has_data ? abd_get_offset(pio->io_abd, pio->io_size - resid) : NULL, lsize, lsize, &zp, zio_write_gang_member_ready, NULL, NULL, zio_write_gang_done, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(has_data); /* * Gang children won't throttle but we should * account for their work, so reserve an allocation * slot for them here. */ VERIFY(metaslab_class_throttle_reserve(mc, zp.zp_copies, cio->io_allocator, cio, flags)); } zio_nowait(cio); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; /* * We didn't allocate this bp, so make sure it doesn't get unmarked. */ pio->io_flags &= ~ZIO_FLAG_FASTWRITE; zio_nowait(zio); return (pio); } /* * The zio_nop_write stage in the pipeline determines if allocating a * new bp is necessary. The nopwrite feature can handle writes in * either syncing or open context (i.e. zil writes) and as a result is * mutually exclusive with dedup. * * By leveraging a cryptographically secure checksum, such as SHA256, we * can compare the checksums of the new data and the old to determine if * allocating a new block is required. Note that our requirements for * cryptographic strength are fairly weak: there can't be any accidental * hash collisions, but we don't need to be secure against intentional * (malicious) collisions. To trigger a nopwrite, you have to be able * to write the file to begin with, and triggering an incorrect (hash * collision) nopwrite is no worse than simply writing to the file. * That said, there are no known attacks against the checksum algorithms * used for nopwrite, assuming that the salt and the checksums * themselves remain secret. */ static zio_t * zio_nop_write(zio_t *zio) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; zio_prop_t *zp = &zio->io_prop; ASSERT(BP_GET_LEVEL(bp) == 0); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(zp->zp_nopwrite); ASSERT(!zp->zp_dedup); ASSERT(zio->io_bp_override == NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Check to see if the original bp and the new bp have matching * characteristics (i.e. same checksum, compression algorithms, etc). * If they don't then just continue with the pipeline which will * allocate a new bp. */ if (BP_IS_HOLE(bp_orig) || !(zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_flags & ZCHECKSUM_FLAG_NOPWRITE) || BP_IS_ENCRYPTED(bp) || BP_IS_ENCRYPTED(bp_orig) || BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) || BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) || BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) || zp->zp_copies != BP_GET_NDVAS(bp_orig)) return (zio); /* * If the checksums match then reset the pipeline so that we * avoid allocating a new bp and issuing any I/O. */ if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) { ASSERT(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_NOPWRITE); ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig)); ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig)); ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF); ASSERT(memcmp(&bp->blk_prop, &bp_orig->blk_prop, sizeof (uint64_t)) == 0); /* * If we're overwriting a block that is currently on an * indirect vdev, then ignore the nopwrite request and * allow a new block to be allocated on a concrete vdev. */ spa_config_enter(zio->io_spa, SCL_VDEV, FTAG, RW_READER); vdev_t *tvd = vdev_lookup_top(zio->io_spa, DVA_GET_VDEV(&bp->blk_dva[0])); if (tvd->vdev_ops == &vdev_indirect_ops) { spa_config_exit(zio->io_spa, SCL_VDEV, FTAG); return (zio); } spa_config_exit(zio->io_spa, SCL_VDEV, FTAG); *bp = *bp_orig; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; zio->io_flags |= ZIO_FLAG_NOPWRITE; } return (zio); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddp = ddt_phys_select(dde, bp); if (zio->io_error == 0) ddt_phys_clear(ddp); /* this ddp doesn't need repair */ if (zio->io_error == 0 && dde->dde_repair_abd == NULL) dde->dde_repair_abd = zio->io_abd; else abd_free(zio->io_abd); mutex_exit(&pio->io_lock); } static zio_t * zio_ddt_read_start(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = ddt_repair_start(ddt, bp); ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *ddp_self = ddt_phys_select(dde, bp); blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (ddp_self == NULL) return (zio); for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp == ddp_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, &blk); zio_nowait(zio_read(zio, zio->io_spa, &blk, abd_alloc_for_io(zio->io_size, B_TRUE), zio->io_size, zio_ddt_child_read_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) | ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark)); } return (zio); } zio_nowait(zio_read(zio, zio->io_spa, bp, zio->io_abd, zio->io_size, NULL, NULL, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark)); return (zio); } static zio_t * zio_ddt_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_DDT_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_vsd; if (ddt == NULL) { ASSERT(spa_load_state(zio->io_spa) != SPA_LOAD_NONE); return (zio); } if (dde == NULL) { zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (NULL); } if (dde->dde_repair_abd != NULL) { abd_copy(zio->io_abd, dde->dde_repair_abd, zio->io_size); zio->io_child_error[ZIO_CHILD_DDT] = 0; } ddt_repair_done(ddt, dde); zio->io_vsd = NULL; } ASSERT(zio->io_vsd == NULL); return (zio); } static boolean_t zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde) { spa_t *spa = zio->io_spa; boolean_t do_raw = !!(zio->io_flags & ZIO_FLAG_RAW); ASSERT(!(zio->io_bp_override && do_raw)); /* * Note: we compare the original data, not the transformed data, * because when zio->io_bp is an override bp, we will not have * pushed the I/O transforms. That's an important optimization * because otherwise we'd compress/encrypt all dmu_sync() data twice. * However, we should never get a raw, override zio so in these * cases we can compare the io_abd directly. This is useful because * it allows us to do dedup verification even if we don't have access * to the original data (for instance, if the encryption keys aren't * loaded). */ for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { zio_t *lio = dde->dde_lead_zio[p]; if (lio != NULL && do_raw) { return (lio->io_size != zio->io_size || abd_cmp(zio->io_abd, lio->io_abd) != 0); } else if (lio != NULL) { return (lio->io_orig_size != zio->io_orig_size || abd_cmp(zio->io_orig_abd, lio->io_orig_abd) != 0); } } for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; if (ddp->ddp_phys_birth != 0 && do_raw) { blkptr_t blk = *zio->io_bp; uint64_t psize; abd_t *tmpabd; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); psize = BP_GET_PSIZE(&blk); if (psize != zio->io_size) return (B_TRUE); ddt_exit(ddt); tmpabd = abd_alloc_for_io(psize, B_TRUE); error = zio_wait(zio_read(NULL, spa, &blk, tmpabd, psize, NULL, NULL, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_RAW, &zio->io_bookmark)); if (error == 0) { if (abd_cmp(tmpabd, zio->io_abd) != 0) error = SET_ERROR(ENOENT); } abd_free(tmpabd); ddt_enter(ddt); return (error != 0); } else if (ddp->ddp_phys_birth != 0) { arc_buf_t *abuf = NULL; arc_flags_t aflags = ARC_FLAG_WAIT; blkptr_t blk = *zio->io_bp; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); if (BP_GET_LSIZE(&blk) != zio->io_orig_size) return (B_TRUE); ddt_exit(ddt); error = arc_read(NULL, spa, &blk, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zio->io_bookmark); if (error == 0) { if (abd_cmp_buf(zio->io_orig_abd, abuf->b_data, zio->io_orig_size) != 0) error = SET_ERROR(ENOENT); arc_buf_destroy(abuf, &abuf); } ddt_enter(ddt); return (error != 0); } } return (B_FALSE); } static void zio_ddt_child_write_ready(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *pio; if (zio->io_error) return; ddt_enter(ddt); ASSERT(dde->dde_lead_zio[p] == zio); ddt_phys_fill(ddp, zio->io_bp); zio_link_t *zl = NULL; while ((pio = zio_walk_parents(zio, &zl)) != NULL) ddt_bp_fill(ddp, pio->io_bp, zio->io_txg); ddt_exit(ddt); } static void zio_ddt_child_write_done(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { zio_link_t *zl = NULL; while (zio_walk_parents(zio, &zl) != NULL) ddt_phys_addref(ddp); } else { ddt_phys_clear(ddp); } ddt_exit(ddt); } static zio_t * zio_ddt_write(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t txg = zio->io_txg; zio_prop_t *zp = &zio->io_prop; int p = zp->zp_copies; zio_t *cio = NULL; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum); ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override); ASSERT(!(zio->io_bp_override && (zio->io_flags & ZIO_FLAG_RAW))); ddt_enter(ddt); dde = ddt_lookup(ddt, bp, B_TRUE); ddp = &dde->dde_phys[p]; if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) { /* * If we're using a weak checksum, upgrade to a strong checksum * and try again. If we're already using a strong checksum, * we can't resolve it, so just convert to an ordinary write. * (And automatically e-mail a paper to Nature?) */ if (!(zio_checksum_table[zp->zp_checksum].ci_flags & ZCHECKSUM_FLAG_DEDUP)) { zp->zp_checksum = spa_dedup_checksum(spa); zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; BP_ZERO(bp); } else { zp->zp_dedup = B_FALSE; BP_SET_DEDUP(bp, B_FALSE); } ASSERT(!BP_GET_DEDUP(bp)); zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (zio); } if (ddp->ddp_phys_birth != 0 || dde->dde_lead_zio[p] != NULL) { if (ddp->ddp_phys_birth != 0) ddt_bp_fill(ddp, bp, txg); if (dde->dde_lead_zio[p] != NULL) zio_add_child(zio, dde->dde_lead_zio[p]); else ddt_phys_addref(ddp); } else if (zio->io_bp_override) { ASSERT(bp->blk_birth == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_fill(ddp, bp); ddt_phys_addref(ddp); } else { cio = zio_write(zio, spa, txg, bp, zio->io_orig_abd, zio->io_orig_size, zio->io_orig_size, zp, zio_ddt_child_write_ready, NULL, NULL, zio_ddt_child_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(cio, zio->io_abd, zio->io_size, 0, NULL); dde->dde_lead_zio[p] = cio; } ddt_exit(ddt); zio_nowait(cio); return (zio); } ddt_entry_t *freedde; /* for debugging */ static zio_t * zio_ddt_free(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ddt_enter(ddt); freedde = dde = ddt_lookup(ddt, bp, B_TRUE); if (dde) { ddp = ddt_phys_select(dde, bp); if (ddp) ddt_phys_decref(ddp); } ddt_exit(ddt); return (zio); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static zio_t * zio_io_to_allocate(spa_t *spa, int allocator) { zio_t *zio; ASSERT(MUTEX_HELD(&spa->spa_allocs[allocator].spaa_lock)); zio = avl_first(&spa->spa_allocs[allocator].spaa_tree); if (zio == NULL) return (NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Try to place a reservation for this zio. If we're unable to * reserve then we throttle. */ ASSERT3U(zio->io_allocator, ==, allocator); if (!metaslab_class_throttle_reserve(zio->io_metaslab_class, zio->io_prop.zp_copies, allocator, zio, 0)) { return (NULL); } avl_remove(&spa->spa_allocs[allocator].spaa_tree, zio); ASSERT3U(zio->io_stage, <, ZIO_STAGE_DVA_ALLOCATE); return (zio); } static zio_t * zio_dva_throttle(zio_t *zio) { spa_t *spa = zio->io_spa; zio_t *nio; metaslab_class_t *mc; /* locate an appropriate allocation class */ mc = spa_preferred_class(spa, zio->io_size, zio->io_prop.zp_type, zio->io_prop.zp_level, zio->io_prop.zp_zpl_smallblk); if (zio->io_priority == ZIO_PRIORITY_SYNC_WRITE || !mc->mc_alloc_throttle_enabled || zio->io_child_type == ZIO_CHILD_GANG || zio->io_flags & ZIO_FLAG_NODATA) { return (zio); } ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); ASSERT3U(zio->io_queued_timestamp, >, 0); ASSERT(zio->io_stage == ZIO_STAGE_DVA_THROTTLE); zbookmark_phys_t *bm = &zio->io_bookmark; /* * We want to try to use as many allocators as possible to help improve * performance, but we also want logically adjacent IOs to be physically * adjacent to improve sequential read performance. We chunk each object * into 2^20 block regions, and then hash based on the objset, object, * level, and region to accomplish both of these goals. */ int allocator = (uint_t)cityhash4(bm->zb_objset, bm->zb_object, bm->zb_level, bm->zb_blkid >> 20) % spa->spa_alloc_count; zio->io_allocator = allocator; zio->io_metaslab_class = mc; mutex_enter(&spa->spa_allocs[allocator].spaa_lock); avl_add(&spa->spa_allocs[allocator].spaa_tree, zio); nio = zio_io_to_allocate(spa, allocator); mutex_exit(&spa->spa_allocs[allocator].spaa_lock); return (nio); } static void zio_allocate_dispatch(spa_t *spa, int allocator) { zio_t *zio; mutex_enter(&spa->spa_allocs[allocator].spaa_lock); zio = zio_io_to_allocate(spa, allocator); mutex_exit(&spa->spa_allocs[allocator].spaa_lock); if (zio == NULL) return; ASSERT3U(zio->io_stage, ==, ZIO_STAGE_DVA_THROTTLE); ASSERT0(zio->io_error); zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_TRUE); } static zio_t * zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc; blkptr_t *bp = zio->io_bp; int error; int flags = 0; if (zio->io_gang_leader == NULL) { ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; } ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_NDVAS(bp)); ASSERT3U(zio->io_prop.zp_copies, >, 0); ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa)); ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp)); flags |= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0; if (zio->io_flags & ZIO_FLAG_NODATA) flags |= METASLAB_DONT_THROTTLE; if (zio->io_flags & ZIO_FLAG_GANG_CHILD) flags |= METASLAB_GANG_CHILD; if (zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE) flags |= METASLAB_ASYNC_ALLOC; /* * if not already chosen, locate an appropriate allocation class */ mc = zio->io_metaslab_class; if (mc == NULL) { mc = spa_preferred_class(spa, zio->io_size, zio->io_prop.zp_type, zio->io_prop.zp_level, zio->io_prop.zp_zpl_smallblk); zio->io_metaslab_class = mc; } /* * Try allocating the block in the usual metaslab class. * If that's full, allocate it in the normal class. * If that's full, allocate as a gang block, * and if all are full, the allocation fails (which shouldn't happen). * * Note that we do not fall back on embedded slog (ZIL) space, to * preserve unfragmented slog space, which is critical for decent * sync write performance. If a log allocation fails, we will fall * back to spa_sync() which is abysmal for performance. */ error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio, zio->io_allocator); /* * Fallback to normal class when an alloc class is full */ if (error == ENOSPC && mc != spa_normal_class(spa)) { /* * If throttling, transfer reservation over to normal class. * The io_allocator slot can remain the same even though we * are switching classes. */ if (mc->mc_alloc_throttle_enabled && (zio->io_flags & ZIO_FLAG_IO_ALLOCATING)) { metaslab_class_throttle_unreserve(mc, zio->io_prop.zp_copies, zio->io_allocator, zio); zio->io_flags &= ~ZIO_FLAG_IO_ALLOCATING; VERIFY(metaslab_class_throttle_reserve( spa_normal_class(spa), zio->io_prop.zp_copies, zio->io_allocator, zio, flags | METASLAB_MUST_RESERVE)); } zio->io_metaslab_class = mc = spa_normal_class(spa); if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) { zfs_dbgmsg("%s: metaslab allocation failure, " "trying normal class: zio %px, size %llu, error %d", spa_name(spa), zio, (u_longlong_t)zio->io_size, error); } error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags, &zio->io_alloc_list, zio, zio->io_allocator); } if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) { if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) { zfs_dbgmsg("%s: metaslab allocation failure, " "trying ganging: zio %px, size %llu, error %d", spa_name(spa), zio, (u_longlong_t)zio->io_size, error); } return (zio_write_gang_block(zio, mc)); } if (error != 0) { if (error != ENOSPC || (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC)) { zfs_dbgmsg("%s: metaslab allocation failure: zio %px, " "size %llu, error %d", spa_name(spa), zio, (u_longlong_t)zio->io_size, error); } zio->io_error = error; } return (zio); } static zio_t * zio_dva_free(zio_t *zio) { metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE); return (zio); } static zio_t * zio_dva_claim(zio_t *zio) { int error; error = metaslab_claim(zio->io_spa, zio->io_bp, zio->io_txg); if (error) zio->io_error = error; return (zio); } /* * Undo an allocation. This is used by zio_done() when an I/O fails * and we want to give back the block we just allocated. * This handles both normal blocks and gang blocks. */ static void zio_dva_unallocate(zio_t *zio, zio_gang_node_t *gn, blkptr_t *bp) { ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp)); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp)) metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE); if (gn != NULL) { for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) { zio_dva_unallocate(zio, gn->gn_child[g], &gn->gn_gbh->zg_blkptr[g]); } } } /* * Try to allocate an intent log block. Return 0 on success, errno on failure. */ int zio_alloc_zil(spa_t *spa, objset_t *os, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t *slog) { int error = 1; zio_alloc_list_t io_alloc_list; ASSERT(txg > spa_syncing_txg(spa)); metaslab_trace_init(&io_alloc_list); /* * Block pointer fields are useful to metaslabs for stats and debugging. * Fill in the obvious ones before calling into metaslab_alloc(). */ BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_PSIZE(new_bp, size); BP_SET_LEVEL(new_bp, 0); /* * When allocating a zil block, we don't have information about * the final destination of the block except the objset it's part * of, so we just hash the objset ID to pick the allocator to get * some parallelism. */ int flags = METASLAB_FASTWRITE | METASLAB_ZIL; int allocator = (uint_t)cityhash4(0, 0, 0, os->os_dsl_dataset->ds_object) % spa->spa_alloc_count; error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, NULL, allocator); *slog = (error == 0); if (error != 0) { error = metaslab_alloc(spa, spa_embedded_log_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, NULL, allocator); } if (error != 0) { error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, flags, &io_alloc_list, NULL, allocator); } metaslab_trace_fini(&io_alloc_list); if (error == 0) { BP_SET_LSIZE(new_bp, size); BP_SET_PSIZE(new_bp, size); BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF); BP_SET_CHECKSUM(new_bp, spa_version(spa) >= SPA_VERSION_SLIM_ZIL ? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG); BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_LEVEL(new_bp, 0); BP_SET_DEDUP(new_bp, 0); BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER); /* * encrypted blocks will require an IV and salt. We generate * these now since we will not be rewriting the bp at * rewrite time. */ if (os->os_encrypted) { uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t salt[ZIO_DATA_SALT_LEN]; BP_SET_CRYPT(new_bp, B_TRUE); VERIFY0(spa_crypt_get_salt(spa, dmu_objset_id(os), salt)); VERIFY0(zio_crypt_generate_iv(iv)); zio_crypt_encode_params_bp(new_bp, salt, iv); } } else { zfs_dbgmsg("%s: zil block allocation failure: " "size %llu, error %d", spa_name(spa), (u_longlong_t)size, error); } return (error); } /* * ========================================================================== * Read and write to physical devices * ========================================================================== */ /* * Issue an I/O to the underlying vdev. Typically the issue pipeline * stops after this stage and will resume upon I/O completion. * However, there are instances where the vdev layer may need to * continue the pipeline when an I/O was not issued. Since the I/O * that was sent to the vdev layer might be different than the one * currently active in the pipeline (see vdev_queue_io()), we explicitly * force the underlying vdev layers to call either zio_execute() or * zio_interrupt() to ensure that the pipeline continues with the correct I/O. */ static zio_t * zio_vdev_io_start(zio_t *zio) { vdev_t *vd = zio->io_vd; uint64_t align; spa_t *spa = zio->io_spa; zio->io_delay = 0; ASSERT(zio->io_error == 0); ASSERT(zio->io_child_error[ZIO_CHILD_VDEV] == 0); if (vd == NULL) { if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_enter(spa, SCL_ZIO, zio, RW_READER); /* * The mirror_ops handle multiple DVAs in a single BP. */ vdev_mirror_ops.vdev_op_io_start(zio); return (NULL); } ASSERT3P(zio->io_logical, !=, zio); if (zio->io_type == ZIO_TYPE_WRITE) { ASSERT(spa->spa_trust_config); /* * Note: the code can handle other kinds of writes, * but we don't expect them. */ if (zio->io_vd->vdev_noalloc) { ASSERT(zio->io_flags & (ZIO_FLAG_PHYSICAL | ZIO_FLAG_SELF_HEAL | ZIO_FLAG_RESILVER | ZIO_FLAG_INDUCE_DAMAGE)); } } align = 1ULL << vd->vdev_top->vdev_ashift; if (!(zio->io_flags & ZIO_FLAG_PHYSICAL) && P2PHASE(zio->io_size, align) != 0) { /* Transform logical writes to be a full physical block size. */ uint64_t asize = P2ROUNDUP(zio->io_size, align); abd_t *abuf = abd_alloc_sametype(zio->io_abd, asize); ASSERT(vd == vd->vdev_top); if (zio->io_type == ZIO_TYPE_WRITE) { abd_copy(abuf, zio->io_abd, zio->io_size); abd_zero_off(abuf, zio->io_size, asize - zio->io_size); } zio_push_transform(zio, abuf, asize, asize, zio_subblock); } /* * If this is not a physical io, make sure that it is properly aligned * before proceeding. */ if (!(zio->io_flags & ZIO_FLAG_PHYSICAL)) { ASSERT0(P2PHASE(zio->io_offset, align)); ASSERT0(P2PHASE(zio->io_size, align)); } else { /* * For physical writes, we allow 512b aligned writes and assume * the device will perform a read-modify-write as necessary. */ ASSERT0(P2PHASE(zio->io_offset, SPA_MINBLOCKSIZE)); ASSERT0(P2PHASE(zio->io_size, SPA_MINBLOCKSIZE)); } VERIFY(zio->io_type != ZIO_TYPE_WRITE || spa_writeable(spa)); /* * If this is a repair I/O, and there's no self-healing involved -- * that is, we're just resilvering what we expect to resilver -- * then don't do the I/O unless zio's txg is actually in vd's DTL. * This prevents spurious resilvering. * * There are a few ways that we can end up creating these spurious * resilver i/os: * * 1. A resilver i/o will be issued if any DVA in the BP has a * dirty DTL. The mirror code will issue resilver writes to * each DVA, including the one(s) that are not on vdevs with dirty * DTLs. * * 2. With nested replication, which happens when we have a * "replacing" or "spare" vdev that's a child of a mirror or raidz. * For example, given mirror(replacing(A+B), C), it's likely that * only A is out of date (it's the new device). In this case, we'll * read from C, then use the data to resilver A+B -- but we don't * actually want to resilver B, just A. The top-level mirror has no * way to know this, so instead we just discard unnecessary repairs * as we work our way down the vdev tree. * * 3. ZTEST also creates mirrors of mirrors, mirrors of raidz, etc. * The same logic applies to any form of nested replication: ditto * + mirror, RAID-Z + replacing, etc. * * However, indirect vdevs point off to other vdevs which may have * DTL's, so we never bypass them. The child i/os on concrete vdevs * will be properly bypassed instead. * * Leaf DTL_PARTIAL can be empty when a legitimate write comes from * a dRAID spare vdev. For example, when a dRAID spare is first * used, its spare blocks need to be written to but the leaf vdev's * of such blocks can have empty DTL_PARTIAL. * * There seemed no clean way to allow such writes while bypassing * spurious ones. At this point, just avoid all bypassing for dRAID * for correctness. */ if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) && !(zio->io_flags & ZIO_FLAG_SELF_HEAL) && zio->io_txg != 0 && /* not a delegated i/o */ vd->vdev_ops != &vdev_indirect_ops && vd->vdev_top->vdev_ops != &vdev_draid_ops && !vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio_vdev_io_bypass(zio); return (zio); } /* * Select the next best leaf I/O to process. Distributed spares are * excluded since they dispatch the I/O directly to a leaf vdev after * applying the dRAID mapping. */ if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops != &vdev_draid_spare_ops && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE || zio->io_type == ZIO_TYPE_TRIM)) { if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio)) return (zio); if ((zio = vdev_queue_io(zio)) == NULL) return (NULL); if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return (NULL); } zio->io_delay = gethrtime(); } vd->vdev_ops->vdev_op_io_start(zio); return (NULL); } static zio_t * zio_vdev_io_done(zio_t *zio) { vdev_t *vd = zio->io_vd; vdev_ops_t *ops = vd ? vd->vdev_ops : &vdev_mirror_ops; boolean_t unexpected_error = B_FALSE; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) { return (NULL); } ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE || zio->io_type == ZIO_TYPE_TRIM); if (zio->io_delay) zio->io_delay = gethrtime() - zio->io_delay; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && vd->vdev_ops != &vdev_draid_spare_ops) { vdev_queue_io_done(zio); if (zio->io_type == ZIO_TYPE_WRITE) vdev_cache_write(zio); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_device_injections(vd, zio, EIO, EILSEQ); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_label_injection(zio, EIO); if (zio->io_error && zio->io_type != ZIO_TYPE_TRIM) { if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); } else { unexpected_error = B_TRUE; } } } ops->vdev_op_io_done(zio); - if (unexpected_error) + if (unexpected_error && vd->vdev_remove_wanted == B_FALSE) VERIFY(vdev_probe(vd, zio) == NULL); return (zio); } /* * This function is used to change the priority of an existing zio that is * currently in-flight. This is used by the arc to upgrade priority in the * event that a demand read is made for a block that is currently queued * as a scrub or async read IO. Otherwise, the high priority read request * would end up having to wait for the lower priority IO. */ void zio_change_priority(zio_t *pio, zio_priority_t priority) { zio_t *cio, *cio_next; zio_link_t *zl = NULL; ASSERT3U(priority, <, ZIO_PRIORITY_NUM_QUEUEABLE); if (pio->io_vd != NULL && pio->io_vd->vdev_ops->vdev_op_leaf) { vdev_queue_change_io_priority(pio, priority); } else { pio->io_priority = priority; } mutex_enter(&pio->io_lock); for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio, &zl); zio_change_priority(cio, priority); } mutex_exit(&pio->io_lock); } /* * For non-raidz ZIOs, we can just copy aside the bad data read from the * disk, and use that to finish the checksum ereport later. */ static void zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr, const abd_t *good_buf) { /* no processing needed */ zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE); } void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr) { void *abd = abd_alloc_sametype(zio->io_abd, zio->io_size); abd_copy(abd, zio->io_abd, zio->io_size); zcr->zcr_cbinfo = zio->io_size; zcr->zcr_cbdata = abd; zcr->zcr_finish = zio_vsd_default_cksum_finish; zcr->zcr_free = zio_abd_free; } static zio_t * zio_vdev_io_assess(zio_t *zio) { vdev_t *vd = zio->io_vd; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) { return (NULL); } if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_exit(zio->io_spa, SCL_ZIO, zio); if (zio->io_vsd != NULL) { zio->io_vsd_ops->vsd_free(zio); zio->io_vsd = NULL; } if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_fault_injection(zio, EIO); /* * If the I/O failed, determine whether we should attempt to retry it. * * On retry, we cut in line in the issue queue, since we don't want * compression/checksumming/etc. work to prevent our (cheap) IO reissue. */ if (zio->io_error && vd == NULL && !(zio->io_flags & (ZIO_FLAG_DONT_RETRY | ZIO_FLAG_IO_RETRY))) { ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */ ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */ zio->io_error = 0; zio->io_flags |= ZIO_FLAG_IO_RETRY | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE; zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, zio_requeue_io_start_cut_in_line); return (NULL); } /* * If we got an error on a leaf device, convert it to ENXIO * if the device is not accessible at all. */ if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf && !vdev_accessible(vd, zio)) zio->io_error = SET_ERROR(ENXIO); /* * If we can't write to an interior vdev (mirror or RAID-Z), * set vdev_cant_write so that we stop trying to allocate from it. */ if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE && vd != NULL && !vd->vdev_ops->vdev_op_leaf) { vdev_dbgmsg(vd, "zio_vdev_io_assess(zio=%px) setting " "cant_write=TRUE due to write failure with ENXIO", zio); vd->vdev_cant_write = B_TRUE; } /* * If a cache flush returns ENOTSUP or ENOTTY, we know that no future * attempts will ever succeed. In this case we set a persistent * boolean flag so that we don't bother with it in the future. */ if ((zio->io_error == ENOTSUP || zio->io_error == ENOTTY) && zio->io_type == ZIO_TYPE_IOCTL && zio->io_cmd == DKIOCFLUSHWRITECACHE && vd != NULL) vd->vdev_nowritecache = B_TRUE; if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && zio->io_physdone != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED)); ASSERT(zio->io_child_type == ZIO_CHILD_VDEV); zio->io_physdone(zio->io_logical); } return (zio); } void zio_vdev_io_reissue(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_stage >>= 1; } void zio_vdev_io_redone(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_DONE); zio->io_stage >>= 1; } void zio_vdev_io_bypass(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_flags |= ZIO_FLAG_IO_BYPASS; zio->io_stage = ZIO_STAGE_VDEV_IO_ASSESS >> 1; } /* * ========================================================================== * Encrypt and store encryption parameters * ========================================================================== */ /* * This function is used for ZIO_STAGE_ENCRYPT. It is responsible for * managing the storage of encryption parameters and passing them to the * lower-level encryption functions. */ static zio_t * zio_encrypt(zio_t *zio) { zio_prop_t *zp = &zio->io_prop; spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t psize = BP_GET_PSIZE(bp); uint64_t dsobj = zio->io_bookmark.zb_objset; dmu_object_type_t ot = BP_GET_TYPE(bp); void *enc_buf = NULL; abd_t *eabd = NULL; uint8_t salt[ZIO_DATA_SALT_LEN]; uint8_t iv[ZIO_DATA_IV_LEN]; uint8_t mac[ZIO_DATA_MAC_LEN]; boolean_t no_crypt = B_FALSE; /* the root zio already encrypted the data */ if (zio->io_child_type == ZIO_CHILD_GANG) return (zio); /* only ZIL blocks are re-encrypted on rewrite */ if (!IO_IS_ALLOCATING(zio) && ot != DMU_OT_INTENT_LOG) return (zio); if (!(zp->zp_encrypt || BP_IS_ENCRYPTED(bp))) { BP_SET_CRYPT(bp, B_FALSE); return (zio); } /* if we are doing raw encryption set the provided encryption params */ if (zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) { ASSERT0(BP_GET_LEVEL(bp)); BP_SET_CRYPT(bp, B_TRUE); BP_SET_BYTEORDER(bp, zp->zp_byteorder); if (ot != DMU_OT_OBJSET) zio_crypt_encode_mac_bp(bp, zp->zp_mac); /* dnode blocks must be written out in the provided byteorder */ if (zp->zp_byteorder != ZFS_HOST_BYTEORDER && ot == DMU_OT_DNODE) { void *bswap_buf = zio_buf_alloc(psize); abd_t *babd = abd_get_from_buf(bswap_buf, psize); ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF); abd_copy_to_buf(bswap_buf, zio->io_abd, psize); dmu_ot_byteswap[DMU_OT_BYTESWAP(ot)].ob_func(bswap_buf, psize); abd_take_ownership_of_buf(babd, B_TRUE); zio_push_transform(zio, babd, psize, psize, NULL); } if (DMU_OT_IS_ENCRYPTED(ot)) zio_crypt_encode_params_bp(bp, zp->zp_salt, zp->zp_iv); return (zio); } /* indirect blocks only maintain a cksum of the lower level MACs */ if (BP_GET_LEVEL(bp) > 0) { BP_SET_CRYPT(bp, B_TRUE); VERIFY0(zio_crypt_do_indirect_mac_checksum_abd(B_TRUE, zio->io_orig_abd, BP_GET_LSIZE(bp), BP_SHOULD_BYTESWAP(bp), mac)); zio_crypt_encode_mac_bp(bp, mac); return (zio); } /* * Objset blocks are a special case since they have 2 256-bit MACs * embedded within them. */ if (ot == DMU_OT_OBJSET) { ASSERT0(DMU_OT_IS_ENCRYPTED(ot)); ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF); BP_SET_CRYPT(bp, B_TRUE); VERIFY0(spa_do_crypt_objset_mac_abd(B_TRUE, spa, dsobj, zio->io_abd, psize, BP_SHOULD_BYTESWAP(bp))); return (zio); } /* unencrypted object types are only authenticated with a MAC */ if (!DMU_OT_IS_ENCRYPTED(ot)) { BP_SET_CRYPT(bp, B_TRUE); VERIFY0(spa_do_crypt_mac_abd(B_TRUE, spa, dsobj, zio->io_abd, psize, mac)); zio_crypt_encode_mac_bp(bp, mac); return (zio); } /* * Later passes of sync-to-convergence may decide to rewrite data * in place to avoid more disk reallocations. This presents a problem * for encryption because this constitutes rewriting the new data with * the same encryption key and IV. However, this only applies to blocks * in the MOS (particularly the spacemaps) and we do not encrypt the * MOS. We assert that the zio is allocating or an intent log write * to enforce this. */ ASSERT(IO_IS_ALLOCATING(zio) || ot == DMU_OT_INTENT_LOG); ASSERT(BP_GET_LEVEL(bp) == 0 || ot == DMU_OT_INTENT_LOG); ASSERT(spa_feature_is_active(spa, SPA_FEATURE_ENCRYPTION)); ASSERT3U(psize, !=, 0); enc_buf = zio_buf_alloc(psize); eabd = abd_get_from_buf(enc_buf, psize); abd_take_ownership_of_buf(eabd, B_TRUE); /* * For an explanation of what encryption parameters are stored * where, see the block comment in zio_crypt.c. */ if (ot == DMU_OT_INTENT_LOG) { zio_crypt_decode_params_bp(bp, salt, iv); } else { BP_SET_CRYPT(bp, B_TRUE); } /* Perform the encryption. This should not fail */ VERIFY0(spa_do_crypt_abd(B_TRUE, spa, &zio->io_bookmark, BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp), salt, iv, mac, psize, zio->io_abd, eabd, &no_crypt)); /* encode encryption metadata into the bp */ if (ot == DMU_OT_INTENT_LOG) { /* * ZIL blocks store the MAC in the embedded checksum, so the * transform must always be applied. */ zio_crypt_encode_mac_zil(enc_buf, mac); zio_push_transform(zio, eabd, psize, psize, NULL); } else { BP_SET_CRYPT(bp, B_TRUE); zio_crypt_encode_params_bp(bp, salt, iv); zio_crypt_encode_mac_bp(bp, mac); if (no_crypt) { ASSERT3U(ot, ==, DMU_OT_DNODE); abd_free(eabd); } else { zio_push_transform(zio, eabd, psize, psize, NULL); } } return (zio); } /* * ========================================================================== * Generate and verify checksums * ========================================================================== */ static zio_t * zio_checksum_generate(zio_t *zio) { blkptr_t *bp = zio->io_bp; enum zio_checksum checksum; if (bp == NULL) { /* * This is zio_write_phys(). * We're either generating a label checksum, or none at all. */ checksum = zio->io_prop.zp_checksum; if (checksum == ZIO_CHECKSUM_OFF) return (zio); ASSERT(checksum == ZIO_CHECKSUM_LABEL); } else { if (BP_IS_GANG(bp) && zio->io_child_type == ZIO_CHILD_GANG) { ASSERT(!IO_IS_ALLOCATING(zio)); checksum = ZIO_CHECKSUM_GANG_HEADER; } else { checksum = BP_GET_CHECKSUM(bp); } } zio_checksum_compute(zio, checksum, zio->io_abd, zio->io_size); return (zio); } static zio_t * zio_checksum_verify(zio_t *zio) { zio_bad_cksum_t info; blkptr_t *bp = zio->io_bp; int error; ASSERT(zio->io_vd != NULL); if (bp == NULL) { /* * This is zio_read_phys(). * We're either verifying a label checksum, or nothing at all. */ if (zio->io_prop.zp_checksum == ZIO_CHECKSUM_OFF) return (zio); ASSERT3U(zio->io_prop.zp_checksum, ==, ZIO_CHECKSUM_LABEL); } if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (error == ECKSUM && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { (void) zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, zio->io_offset, zio->io_size, &info); mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_checksum_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); } } return (zio); } /* * Called by RAID-Z to ensure we don't compute the checksum twice. */ void zio_checksum_verified(zio_t *zio) { zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } /* * ========================================================================== * Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other. * An error of 0 indicates success. ENXIO indicates whole-device failure, * which may be transient (e.g. unplugged) or permanent. ECKSUM and EIO * indicate errors that are specific to one I/O, and most likely permanent. * Any other error is presumed to be worse because we weren't expecting it. * ========================================================================== */ int zio_worst_error(int e1, int e2) { static int zio_error_rank[] = { 0, ENXIO, ECKSUM, EIO }; int r1, r2; for (r1 = 0; r1 < sizeof (zio_error_rank) / sizeof (int); r1++) if (e1 == zio_error_rank[r1]) break; for (r2 = 0; r2 < sizeof (zio_error_rank) / sizeof (int); r2++) if (e2 == zio_error_rank[r2]) break; return (r1 > r2 ? e1 : e2); } /* * ========================================================================== * I/O completion * ========================================================================== */ static zio_t * zio_ready(zio_t *zio) { blkptr_t *bp = zio->io_bp; zio_t *pio, *pio_next; zio_link_t *zl = NULL; if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT | ZIO_CHILD_DDT_BIT, ZIO_WAIT_READY)) { return (NULL); } if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp) || (zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0); zio->io_ready(zio); } if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; if (zio->io_error != 0) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_metaslab_class != NULL); /* * We were unable to allocate anything, unreserve and * issue the next I/O to allocate. */ metaslab_class_throttle_unreserve( zio->io_metaslab_class, zio->io_prop.zp_copies, zio->io_allocator, zio); zio_allocate_dispatch(zio->io_spa, zio->io_allocator); } } mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_READY] = 1; pio = zio_walk_parents(zio, &zl); mutex_exit(&zio->io_lock); /* * As we notify zio's parents, new parents could be added. * New parents go to the head of zio's io_parent_list, however, * so we will (correctly) not notify them. The remainder of zio's * io_parent_list, from 'pio_next' onward, cannot change because * all parents must wait for us to be done before they can be done. */ for (; pio != NULL; pio = pio_next) { pio_next = zio_walk_parents(zio, &zl); zio_notify_parent(pio, zio, ZIO_WAIT_READY, NULL); } if (zio->io_flags & ZIO_FLAG_NODATA) { if (BP_IS_GANG(bp)) { zio->io_flags &= ~ZIO_FLAG_NODATA; } else { ASSERT((uintptr_t)zio->io_abd < SPA_MAXBLOCKSIZE); zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } } if (zio_injection_enabled && zio->io_spa->spa_syncing_txg == zio->io_txg) zio_handle_ignored_writes(zio); return (zio); } /* * Update the allocation throttle accounting. */ static void zio_dva_throttle_done(zio_t *zio) { zio_t *lio __maybe_unused = zio->io_logical; zio_t *pio = zio_unique_parent(zio); vdev_t *vd = zio->io_vd; int flags = METASLAB_ASYNC_ALLOC; ASSERT3P(zio->io_bp, !=, NULL); ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE); ASSERT3U(zio->io_priority, ==, ZIO_PRIORITY_ASYNC_WRITE); ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV); ASSERT(vd != NULL); ASSERT3P(vd, ==, vd->vdev_top); ASSERT(zio_injection_enabled || !(zio->io_flags & ZIO_FLAG_IO_RETRY)); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT(zio->io_flags & ZIO_FLAG_IO_ALLOCATING); ASSERT(!(lio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(!(lio->io_orig_flags & ZIO_FLAG_NODATA)); /* * Parents of gang children can have two flavors -- ones that * allocated the gang header (will have ZIO_FLAG_IO_REWRITE set) * and ones that allocated the constituent blocks. The allocation * throttle needs to know the allocating parent zio so we must find * it here. */ if (pio->io_child_type == ZIO_CHILD_GANG) { /* * If our parent is a rewrite gang child then our grandparent * would have been the one that performed the allocation. */ if (pio->io_flags & ZIO_FLAG_IO_REWRITE) pio = zio_unique_parent(pio); flags |= METASLAB_GANG_CHILD; } ASSERT(IO_IS_ALLOCATING(pio)); ASSERT3P(zio, !=, zio->io_logical); ASSERT(zio->io_logical != NULL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR)); ASSERT0(zio->io_flags & ZIO_FLAG_NOPWRITE); ASSERT(zio->io_metaslab_class != NULL); mutex_enter(&pio->io_lock); metaslab_group_alloc_decrement(zio->io_spa, vd->vdev_id, pio, flags, pio->io_allocator, B_TRUE); mutex_exit(&pio->io_lock); metaslab_class_throttle_unreserve(zio->io_metaslab_class, 1, pio->io_allocator, pio); /* * Call into the pipeline to see if there is more work that * needs to be done. If there is work to be done it will be * dispatched to another taskq thread. */ zio_allocate_dispatch(zio->io_spa, pio->io_allocator); } static zio_t * zio_done(zio_t *zio) { /* * Always attempt to keep stack usage minimal here since * we can be called recursively up to 19 levels deep. */ const uint64_t psize = zio->io_size; zio_t *pio, *pio_next; zio_link_t *zl = NULL; /* * If our children haven't all completed, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_ALL_BITS, ZIO_WAIT_DONE)) { return (NULL); } /* * If the allocation throttle is enabled, then update the accounting. * We only track child I/Os that are part of an allocating async * write. We must do this since the allocation is performed * by the logical I/O but the actual write is done by child I/Os. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING && zio->io_child_type == ZIO_CHILD_VDEV) { ASSERT(zio->io_metaslab_class != NULL); ASSERT(zio->io_metaslab_class->mc_alloc_throttle_enabled); zio_dva_throttle_done(zio); } /* * If the allocation throttle is enabled, verify that * we have decremented the refcounts for every I/O that was throttled. */ if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE); ASSERT(zio->io_bp != NULL); metaslab_group_alloc_verify(zio->io_spa, zio->io_bp, zio, zio->io_allocator); VERIFY(zfs_refcount_not_held(&zio->io_metaslab_class-> mc_allocator[zio->io_allocator].mca_alloc_slots, zio)); } for (int c = 0; c < ZIO_CHILD_TYPES; c++) for (int w = 0; w < ZIO_WAIT_TYPES; w++) ASSERT(zio->io_children[c][w] == 0); if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) { ASSERT(zio->io_bp->blk_pad[0] == 0); ASSERT(zio->io_bp->blk_pad[1] == 0); ASSERT(memcmp(zio->io_bp, &zio->io_bp_copy, sizeof (blkptr_t)) == 0 || (zio->io_bp == zio_unique_parent(zio)->io_bp)); if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) && zio->io_bp_override == NULL && !(zio->io_flags & ZIO_FLAG_IO_REPAIR)) { ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 || (BP_COUNT_GANG(zio->io_bp) == BP_GET_NDVAS(zio->io_bp))); } if (zio->io_flags & ZIO_FLAG_NOPWRITE) VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } /* * If there were child vdev/gang/ddt errors, they apply to us now. */ zio_inherit_child_errors(zio, ZIO_CHILD_VDEV); zio_inherit_child_errors(zio, ZIO_CHILD_GANG); zio_inherit_child_errors(zio, ZIO_CHILD_DDT); /* * If the I/O on the transformed data was successful, generate any * checksum reports now while we still have the transformed data. */ if (zio->io_error == 0) { while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; uint64_t align = zcr->zcr_align; uint64_t asize = P2ROUNDUP(psize, align); abd_t *adata = zio->io_abd; if (adata != NULL && asize != psize) { adata = abd_alloc(asize, B_TRUE); abd_copy(adata, zio->io_abd, psize); abd_zero_off(adata, psize, asize - psize); } zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, adata); zfs_ereport_free_checksum(zcr); if (adata != NULL && asize != psize) abd_free(adata); } } zio_pop_transforms(zio); /* note: may set zio->io_error */ vdev_stat_update(zio, psize); /* * If this I/O is attached to a particular vdev is slow, exceeding * 30 seconds to complete, post an error described the I/O delay. * We ignore these errors if the device is currently unavailable. */ if (zio->io_delay >= MSEC2NSEC(zio_slow_io_ms)) { if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) { /* * We want to only increment our slow IO counters if * the IO is valid (i.e. not if the drive is removed). * * zfs_ereport_post() will also do these checks, but * it can also ratelimit and have other failures, so we * need to increment the slow_io counters independent * of it. */ if (zfs_ereport_is_valid(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, zio)) { mutex_enter(&zio->io_vd->vdev_stat_lock); zio->io_vd->vdev_stat.vs_slow_ios++; mutex_exit(&zio->io_vd->vdev_stat_lock); (void) zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); } } } if (zio->io_error) { /* * If this I/O is attached to a particular vdev, * generate an error message describing the I/O failure * at the block level. We ignore these errors if the * device is currently unavailable. */ if (zio->io_error != ECKSUM && zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) { int ret = zfs_ereport_post(FM_EREPORT_ZFS_IO, zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0); if (ret != EALREADY) { mutex_enter(&zio->io_vd->vdev_stat_lock); if (zio->io_type == ZIO_TYPE_READ) zio->io_vd->vdev_stat.vs_read_errors++; else if (zio->io_type == ZIO_TYPE_WRITE) zio->io_vd->vdev_stat.vs_write_errors++; mutex_exit(&zio->io_vd->vdev_stat_lock); } } if ((zio->io_error == EIO || !(zio->io_flags & (ZIO_FLAG_SPECULATIVE | ZIO_FLAG_DONT_PROPAGATE))) && zio == zio->io_logical) { /* * For logical I/O requests, tell the SPA to log the * error and generate a logical data ereport. */ spa_log_error(zio->io_spa, &zio->io_bookmark); (void) zfs_ereport_post(FM_EREPORT_ZFS_DATA, zio->io_spa, NULL, &zio->io_bookmark, zio, 0); } } if (zio->io_error && zio == zio->io_logical) { /* * Determine whether zio should be reexecuted. This will * propagate all the way to the root via zio_notify_parent(). */ ASSERT(zio->io_vd == NULL && zio->io_bp != NULL); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (IO_IS_ALLOCATING(zio) && !(zio->io_flags & ZIO_FLAG_CANFAIL)) { if (zio->io_error != ENOSPC) zio->io_reexecute |= ZIO_REEXECUTE_NOW; else zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; } if ((zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_FREE) && !(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_error == ENXIO && spa_load_state(zio->io_spa) == SPA_LOAD_NONE && spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; /* * Here is a possibly good place to attempt to do * either combinatorial reconstruction or error correction * based on checksums. It also might be a good place * to send out preliminary ereports before we suspend * processing. */ } /* * If there were logical child errors, they apply to us now. * We defer this until now to avoid conflating logical child * errors with errors that happened to the zio itself when * updating vdev stats and reporting FMA events above. */ zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL); if ((zio->io_error || zio->io_reexecute) && IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio && !(zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE))) zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp); zio_gang_tree_free(&zio->io_gang_tree); /* * Godfather I/Os should never suspend. */ if ((zio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) zio->io_reexecute &= ~ZIO_REEXECUTE_SUSPEND; if (zio->io_reexecute) { /* * This is a logical I/O that wants to reexecute. * * Reexecute is top-down. When an i/o fails, if it's not * the root, it simply notifies its parent and sticks around. * The parent, seeing that it still has children in zio_done(), * does the same. This percolates all the way up to the root. * The root i/o will reexecute or suspend the entire tree. * * This approach ensures that zio_reexecute() honors * all the original i/o dependency relationships, e.g. * parents not executing until children are ready. */ ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); zio->io_gang_leader = NULL; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * "The Godfather" I/O monitors its children but is * not a true parent to them. It will track them through * the pipeline but severs its ties whenever they get into * trouble (e.g. suspended). This allows "The Godfather" * I/O to return status without blocking. */ zl = NULL; for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) { zio_link_t *remove_zl = zl; pio_next = zio_walk_parents(zio, &zl); if ((pio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) { zio_remove_child(pio, zio, remove_zl); /* * This is a rare code path, so we don't * bother with "next_to_execute". */ zio_notify_parent(pio, zio, ZIO_WAIT_DONE, NULL); } } if ((pio = zio_unique_parent(zio)) != NULL) { /* * We're not a root i/o, so there's nothing to do * but notify our parent. Don't propagate errors * upward since we haven't permanently failed yet. */ ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); zio->io_flags |= ZIO_FLAG_DONT_PROPAGATE; /* * This is a rare code path, so we don't bother with * "next_to_execute". */ zio_notify_parent(pio, zio, ZIO_WAIT_DONE, NULL); } else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) { /* * We'd fail again if we reexecuted now, so suspend * until conditions improve (e.g. device comes online). */ zio_suspend(zio->io_spa, zio, ZIO_SUSPEND_IOERR); } else { /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, zio_reexecute, zio, 0, &zio->io_tqent); } return (NULL); } ASSERT(zio->io_child_count == 0); ASSERT(zio->io_reexecute == 0); ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL)); /* * Report any checksum errors, since the I/O is complete. */ while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, NULL); zfs_ereport_free_checksum(zcr); } if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp && !BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) && !(zio->io_flags & ZIO_FLAG_NOPWRITE)) { metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp); } /* * It is the responsibility of the done callback to ensure that this * particular zio is no longer discoverable for adoption, and as * such, cannot acquire any new parents. */ if (zio->io_done) zio->io_done(zio); mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * We are done executing this zio. We may want to execute a parent * next. See the comment in zio_notify_parent(). */ zio_t *next_to_execute = NULL; zl = NULL; for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) { zio_link_t *remove_zl = zl; pio_next = zio_walk_parents(zio, &zl); zio_remove_child(pio, zio, remove_zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE, &next_to_execute); } if (zio->io_waiter != NULL) { mutex_enter(&zio->io_lock); zio->io_executor = NULL; cv_broadcast(&zio->io_cv); mutex_exit(&zio->io_lock); } else { zio_destroy(zio); } return (next_to_execute); } /* * ========================================================================== * I/O pipeline definition * ========================================================================== */ static zio_pipe_stage_t *zio_pipeline[] = { NULL, zio_read_bp_init, zio_write_bp_init, zio_free_bp_init, zio_issue_async, zio_write_compress, zio_encrypt, zio_checksum_generate, zio_nop_write, zio_ddt_read_start, zio_ddt_read_done, zio_ddt_write, zio_ddt_free, zio_gang_assemble, zio_gang_issue, zio_dva_throttle, zio_dva_allocate, zio_dva_free, zio_dva_claim, zio_ready, zio_vdev_io_start, zio_vdev_io_done, zio_vdev_io_assess, zio_checksum_verify, zio_done }; /* * Compare two zbookmark_phys_t's to see which we would reach first in a * pre-order traversal of the object tree. * * This is simple in every case aside from the meta-dnode object. For all other * objects, we traverse them in order (object 1 before object 2, and so on). * However, all of these objects are traversed while traversing object 0, since * the data it points to is the list of objects. Thus, we need to convert to a * canonical representation so we can compare meta-dnode bookmarks to * non-meta-dnode bookmarks. * * We do this by calculating "equivalents" for each field of the zbookmark. * zbookmarks outside of the meta-dnode use their own object and level, and * calculate the level 0 equivalent (the first L0 blkid that is contained in the * blocks this bookmark refers to) by multiplying their blkid by their span * (the number of L0 blocks contained within one block at their level). * zbookmarks inside the meta-dnode calculate their object equivalent * (which is L0equiv * dnodes per data block), use 0 for their L0equiv, and use * level + 1<<31 (any value larger than a level could ever be) for their level. * This causes them to always compare before a bookmark in their object * equivalent, compare appropriately to bookmarks in other objects, and to * compare appropriately to other bookmarks in the meta-dnode. */ int zbookmark_compare(uint16_t dbss1, uint8_t ibs1, uint16_t dbss2, uint8_t ibs2, const zbookmark_phys_t *zb1, const zbookmark_phys_t *zb2) { /* * These variables represent the "equivalent" values for the zbookmark, * after converting zbookmarks inside the meta dnode to their * normal-object equivalents. */ uint64_t zb1obj, zb2obj; uint64_t zb1L0, zb2L0; uint64_t zb1level, zb2level; if (zb1->zb_object == zb2->zb_object && zb1->zb_level == zb2->zb_level && zb1->zb_blkid == zb2->zb_blkid) return (0); IMPLY(zb1->zb_level > 0, ibs1 >= SPA_MINBLOCKSHIFT); IMPLY(zb2->zb_level > 0, ibs2 >= SPA_MINBLOCKSHIFT); /* * BP_SPANB calculates the span in blocks. */ zb1L0 = (zb1->zb_blkid) * BP_SPANB(ibs1, zb1->zb_level); zb2L0 = (zb2->zb_blkid) * BP_SPANB(ibs2, zb2->zb_level); if (zb1->zb_object == DMU_META_DNODE_OBJECT) { zb1obj = zb1L0 * (dbss1 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb1L0 = 0; zb1level = zb1->zb_level + COMPARE_META_LEVEL; } else { zb1obj = zb1->zb_object; zb1level = zb1->zb_level; } if (zb2->zb_object == DMU_META_DNODE_OBJECT) { zb2obj = zb2L0 * (dbss2 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT)); zb2L0 = 0; zb2level = zb2->zb_level + COMPARE_META_LEVEL; } else { zb2obj = zb2->zb_object; zb2level = zb2->zb_level; } /* Now that we have a canonical representation, do the comparison. */ if (zb1obj != zb2obj) return (zb1obj < zb2obj ? -1 : 1); else if (zb1L0 != zb2L0) return (zb1L0 < zb2L0 ? -1 : 1); else if (zb1level != zb2level) return (zb1level > zb2level ? -1 : 1); /* * This can (theoretically) happen if the bookmarks have the same object * and level, but different blkids, if the block sizes are not the same. * There is presently no way to change the indirect block sizes */ return (0); } /* * This function checks the following: given that last_block is the place that * our traversal stopped last time, does that guarantee that we've visited * every node under subtree_root? Therefore, we can't just use the raw output * of zbookmark_compare. We have to pass in a modified version of * subtree_root; by incrementing the block id, and then checking whether * last_block is before or equal to that, we can tell whether or not having * visited last_block implies that all of subtree_root's children have been * visited. */ boolean_t zbookmark_subtree_completed(const dnode_phys_t *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block) { zbookmark_phys_t mod_zb = *subtree_root; mod_zb.zb_blkid++; ASSERT0(last_block->zb_level); /* The objset_phys_t isn't before anything. */ if (dnp == NULL) return (B_FALSE); /* * We pass in 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT) for the * data block size in sectors, because that variable is only used if * the bookmark refers to a block in the meta-dnode. Since we don't * know without examining it what object it refers to, and there's no * harm in passing in this value in other cases, we always pass it in. * * We pass in 0 for the indirect block size shift because zb2 must be * level 0. The indirect block size is only used to calculate the span * of the bookmark, but since the bookmark must be level 0, the span is * always 1, so the math works out. * * If you make changes to how the zbookmark_compare code works, be sure * to make sure that this code still works afterwards. */ return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift, 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, &mod_zb, last_block) <= 0); } /* * This function is similar to zbookmark_subtree_completed(), but returns true * if subtree_root is equal or ahead of last_block, i.e. still to be done. */ boolean_t zbookmark_subtree_tbd(const dnode_phys_t *dnp, const zbookmark_phys_t *subtree_root, const zbookmark_phys_t *last_block) { ASSERT0(last_block->zb_level); if (dnp == NULL) return (B_FALSE); return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift, 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, subtree_root, last_block) >= 0); } EXPORT_SYMBOL(zio_type_name); EXPORT_SYMBOL(zio_buf_alloc); EXPORT_SYMBOL(zio_data_buf_alloc); EXPORT_SYMBOL(zio_buf_free); EXPORT_SYMBOL(zio_data_buf_free); ZFS_MODULE_PARAM(zfs_zio, zio_, slow_io_ms, INT, ZMOD_RW, "Max I/O completion time (milliseconds) before marking it as slow"); ZFS_MODULE_PARAM(zfs_zio, zio_, requeue_io_start_cut_in_line, INT, ZMOD_RW, "Prioritize requeued I/O"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_deferred_free, UINT, ZMOD_RW, "Defer frees starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_dont_compress, UINT, ZMOD_RW, "Don't compress starting in this pass"); ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_rewrite, UINT, ZMOD_RW, "Rewrite new bps starting in this pass"); ZFS_MODULE_PARAM(zfs_zio, zio_, dva_throttle_enabled, INT, ZMOD_RW, "Throttle block allocations in the ZIO pipeline"); ZFS_MODULE_PARAM(zfs_zio, zio_, deadman_log_all, INT, ZMOD_RW, "Log all slow ZIOs, not just those with vdevs"); diff --git a/tests/zfs-tests/include/libtest.shlib b/tests/zfs-tests/include/libtest.shlib index d163fc7c8ccc..02e6a500a71a 100644 --- a/tests/zfs-tests/include/libtest.shlib +++ b/tests/zfs-tests/include/libtest.shlib @@ -1,3849 +1,3849 @@ # # CDDL HEADER START # # The contents of this file are subject to the terms of the # Common Development and Distribution License (the "License"). # You may not use this file except in compliance with the License. # # You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE # or https://opensource.org/licenses/CDDL-1.0. # See the License for the specific language governing permissions # and limitations under the License. # # When distributing Covered Code, include this CDDL HEADER in each # file and include the License file at usr/src/OPENSOLARIS.LICENSE. # If applicable, add the following below this CDDL HEADER, with the # fields enclosed by brackets "[]" replaced with your own identifying # information: Portions Copyright [yyyy] [name of copyright owner] # # CDDL HEADER END # # # Copyright (c) 2009, Sun Microsystems Inc. All rights reserved. # Copyright (c) 2012, 2020, Delphix. All rights reserved. # Copyright (c) 2017, Tim Chase. All rights reserved. # Copyright (c) 2017, Nexenta Systems Inc. All rights reserved. # Copyright (c) 2017, Lawrence Livermore National Security LLC. # Copyright (c) 2017, Datto Inc. All rights reserved. # Copyright (c) 2017, Open-E Inc. All rights reserved. # Copyright (c) 2021, The FreeBSD Foundation. # Use is subject to license terms. # . ${STF_SUITE}/include/tunables.cfg . ${STF_TOOLS}/include/logapi.shlib . ${STF_SUITE}/include/math.shlib . ${STF_SUITE}/include/blkdev.shlib # # Apply constrained path when available. This is required since the # PATH may have been modified by sudo's secure_path behavior. # if [ -n "$STF_PATH" ]; then export PATH="$STF_PATH" fi # # Generic dot version comparison function # # Returns success when version $1 is greater than or equal to $2. # function compare_version_gte { [ "$(printf "$1\n$2" | sort -V | tail -n1)" = "$1" ] } # Linux kernel version comparison function # # $1 Linux version ("4.10", "2.6.32") or blank for installed Linux version # # Used for comparison: if [ $(linux_version) -ge $(linux_version "2.6.32") ] # function linux_version { typeset ver="$1" [ -z "$ver" ] && ver=$(uname -r | grep -Eo "^[0-9]+\.[0-9]+\.[0-9]+") typeset version major minor _ IFS='.' read -r version major minor _ <<<"$ver" [ -z "$version" ] && version=0 [ -z "$major" ] && major=0 [ -z "$minor" ] && minor=0 echo $((version * 100000 + major * 1000 + minor)) } # Determine if this is a Linux test system # # Return 0 if platform Linux, 1 if otherwise function is_linux { [ "$UNAME" = "Linux" ] } # Determine if this is an illumos test system # # Return 0 if platform illumos, 1 if otherwise function is_illumos { [ "$UNAME" = "illumos" ] } # Determine if this is a FreeBSD test system # # Return 0 if platform FreeBSD, 1 if otherwise function is_freebsd { [ "$UNAME" = "FreeBSD" ] } # Determine if this is a 32-bit system # # Return 0 if platform is 32-bit, 1 if otherwise function is_32bit { [ $(getconf LONG_BIT) = "32" ] } # Determine if kmemleak is enabled # # Return 0 if kmemleak is enabled, 1 if otherwise function is_kmemleak { is_linux && [ -e /sys/kernel/debug/kmemleak ] } # Determine whether a dataset is mounted # # $1 dataset name # $2 filesystem type; optional - defaulted to zfs # # Return 0 if dataset is mounted; 1 if unmounted; 2 on error function ismounted { typeset fstype=$2 [[ -z $fstype ]] && fstype=zfs typeset out dir name case $fstype in zfs) if [[ "$1" == "/"* ]] ; then ! zfs mount | awk -v fs="$1" '$2 == fs {exit 1}' else ! zfs mount | awk -v ds="$1" '$1 == ds {exit 1}' fi ;; ufs|nfs) if is_freebsd; then mount -pt $fstype | while read dev dir _t _flags; do [[ "$1" == "$dev" || "$1" == "$dir" ]] && return 0 done else out=$(df -F $fstype $1 2>/dev/null) || return dir=${out%%\(*} dir=${dir%% *} name=${out##*\(} name=${name%%\)*} name=${name%% *} [[ "$1" == "$dir" || "$1" == "$name" ]] && return 0 fi ;; ext*) df -t $fstype $1 > /dev/null 2>&1 ;; zvol) if [[ -L "$ZVOL_DEVDIR/$1" ]]; then link=$(readlink -f $ZVOL_DEVDIR/$1) [[ -n "$link" ]] && \ mount | grep -q "^$link" && \ return 0 fi ;; *) false ;; esac } # Return 0 if a dataset is mounted; 1 otherwise # # $1 dataset name # $2 filesystem type; optional - defaulted to zfs function mounted { ismounted $1 $2 } # Return 0 if a dataset is unmounted; 1 otherwise # # $1 dataset name # $2 filesystem type; optional - defaulted to zfs function unmounted { ! ismounted $1 $2 } function default_setup { default_setup_noexit "$@" log_pass } function default_setup_no_mountpoint { default_setup_noexit "$1" "$2" "$3" "yes" log_pass } # # Given a list of disks, setup storage pools and datasets. # function default_setup_noexit { typeset disklist=$1 typeset container=$2 typeset volume=$3 typeset no_mountpoint=$4 log_note begin default_setup_noexit if is_global_zone; then if poolexists $TESTPOOL ; then destroy_pool $TESTPOOL fi [[ -d /$TESTPOOL ]] && rm -rf /$TESTPOOL log_must zpool create -f $TESTPOOL $disklist else reexport_pool fi rm -rf $TESTDIR || log_unresolved Could not remove $TESTDIR mkdir -p $TESTDIR || log_unresolved Could not create $TESTDIR log_must zfs create $TESTPOOL/$TESTFS if [[ -z $no_mountpoint ]]; then log_must zfs set mountpoint=$TESTDIR $TESTPOOL/$TESTFS fi if [[ -n $container ]]; then rm -rf $TESTDIR1 || \ log_unresolved Could not remove $TESTDIR1 mkdir -p $TESTDIR1 || \ log_unresolved Could not create $TESTDIR1 log_must zfs create $TESTPOOL/$TESTCTR log_must zfs set canmount=off $TESTPOOL/$TESTCTR log_must zfs create $TESTPOOL/$TESTCTR/$TESTFS1 if [[ -z $no_mountpoint ]]; then log_must zfs set mountpoint=$TESTDIR1 \ $TESTPOOL/$TESTCTR/$TESTFS1 fi fi if [[ -n $volume ]]; then if is_global_zone ; then log_must zfs create -V $VOLSIZE $TESTPOOL/$TESTVOL block_device_wait else log_must zfs create $TESTPOOL/$TESTVOL fi fi } # # Given a list of disks, setup a storage pool, file system and # a container. # function default_container_setup { typeset disklist=$1 default_setup "$disklist" "true" } # # Given a list of disks, setup a storage pool,file system # and a volume. # function default_volume_setup { typeset disklist=$1 default_setup "$disklist" "" "true" } # # Given a list of disks, setup a storage pool,file system, # a container and a volume. # function default_container_volume_setup { typeset disklist=$1 default_setup "$disklist" "true" "true" } # # Create a snapshot on a filesystem or volume. Defaultly create a snapshot on # filesystem # # $1 Existing filesystem or volume name. Default, $TESTPOOL/$TESTFS # $2 snapshot name. Default, $TESTSNAP # function create_snapshot { typeset fs_vol=${1:-$TESTPOOL/$TESTFS} typeset snap=${2:-$TESTSNAP} [[ -z $fs_vol ]] && log_fail "Filesystem or volume's name is undefined." [[ -z $snap ]] && log_fail "Snapshot's name is undefined." if snapexists $fs_vol@$snap; then log_fail "$fs_vol@$snap already exists." fi datasetexists $fs_vol || \ log_fail "$fs_vol must exist." log_must zfs snapshot $fs_vol@$snap } # # Create a clone from a snapshot, default clone name is $TESTCLONE. # # $1 Existing snapshot, $TESTPOOL/$TESTFS@$TESTSNAP is default. # $2 Clone name, $TESTPOOL/$TESTCLONE is default. # function create_clone # snapshot clone { typeset snap=${1:-$TESTPOOL/$TESTFS@$TESTSNAP} typeset clone=${2:-$TESTPOOL/$TESTCLONE} [[ -z $snap ]] && \ log_fail "Snapshot name is undefined." [[ -z $clone ]] && \ log_fail "Clone name is undefined." log_must zfs clone $snap $clone } # # Create a bookmark of the given snapshot. Defaultly create a bookmark on # filesystem. # # $1 Existing filesystem or volume name. Default, $TESTFS # $2 Existing snapshot name. Default, $TESTSNAP # $3 bookmark name. Default, $TESTBKMARK # function create_bookmark { typeset fs_vol=${1:-$TESTFS} typeset snap=${2:-$TESTSNAP} typeset bkmark=${3:-$TESTBKMARK} [[ -z $fs_vol ]] && log_fail "Filesystem or volume's name is undefined." [[ -z $snap ]] && log_fail "Snapshot's name is undefined." [[ -z $bkmark ]] && log_fail "Bookmark's name is undefined." if bkmarkexists $fs_vol#$bkmark; then log_fail "$fs_vol#$bkmark already exists." fi datasetexists $fs_vol || \ log_fail "$fs_vol must exist." snapexists $fs_vol@$snap || \ log_fail "$fs_vol@$snap must exist." log_must zfs bookmark $fs_vol@$snap $fs_vol#$bkmark } # # Create a temporary clone result of an interrupted resumable 'zfs receive' # $1 Destination filesystem name. Must not exist, will be created as the result # of this function along with its %recv temporary clone # $2 Source filesystem name. Must not exist, will be created and destroyed # function create_recv_clone { typeset recvfs="$1" typeset sendfs="${2:-$TESTPOOL/create_recv_clone}" typeset snap="$sendfs@snap1" typeset incr="$sendfs@snap2" typeset mountpoint="$TESTDIR/create_recv_clone" typeset sendfile="$TESTDIR/create_recv_clone.zsnap" [[ -z $recvfs ]] && log_fail "Recv filesystem's name is undefined." datasetexists $recvfs && log_fail "Recv filesystem must not exist." datasetexists $sendfs && log_fail "Send filesystem must not exist." log_must zfs create -o compression=off -o mountpoint="$mountpoint" $sendfs log_must zfs snapshot $snap log_must eval "zfs send $snap | zfs recv -u $recvfs" log_must mkfile 1m "$mountpoint/data" log_must zfs snapshot $incr log_must eval "zfs send -i $snap $incr | dd bs=10K count=1 \ iflag=fullblock > $sendfile" log_mustnot eval "zfs recv -su $recvfs < $sendfile" destroy_dataset "$sendfs" "-r" log_must rm -f "$sendfile" if [[ $(get_prop 'inconsistent' "$recvfs/%recv") -ne 1 ]]; then log_fail "Error creating temporary $recvfs/%recv clone" fi } function default_mirror_setup { default_mirror_setup_noexit $1 $2 $3 log_pass } # # Given a pair of disks, set up a storage pool and dataset for the mirror # @parameters: $1 the primary side of the mirror # $2 the secondary side of the mirror # @uses: ZPOOL ZFS TESTPOOL TESTFS function default_mirror_setup_noexit { readonly func="default_mirror_setup_noexit" typeset primary=$1 typeset secondary=$2 [[ -z $primary ]] && \ log_fail "$func: No parameters passed" [[ -z $secondary ]] && \ log_fail "$func: No secondary partition passed" [[ -d /$TESTPOOL ]] && rm -rf /$TESTPOOL log_must zpool create -f $TESTPOOL mirror $@ log_must zfs create $TESTPOOL/$TESTFS log_must zfs set mountpoint=$TESTDIR $TESTPOOL/$TESTFS } # # Destroy the configured testpool mirrors. # the mirrors are of the form ${TESTPOOL}{number} # @uses: ZPOOL ZFS TESTPOOL function destroy_mirrors { default_cleanup_noexit log_pass } function default_raidz_setup { default_raidz_setup_noexit "$*" log_pass } # # Given a minimum of two disks, set up a storage pool and dataset for the raid-z # $1 the list of disks # function default_raidz_setup_noexit { typeset disklist="$*" disks=(${disklist[*]}) if [[ ${#disks[*]} -lt 2 ]]; then log_fail "A raid-z requires a minimum of two disks." fi [[ -d /$TESTPOOL ]] && rm -rf /$TESTPOOL log_must zpool create -f $TESTPOOL raidz $disklist log_must zfs create $TESTPOOL/$TESTFS log_must zfs set mountpoint=$TESTDIR $TESTPOOL/$TESTFS } # # Common function used to cleanup storage pools and datasets. # # Invoked at the start of the test suite to ensure the system # is in a known state, and also at the end of each set of # sub-tests to ensure errors from one set of tests doesn't # impact the execution of the next set. function default_cleanup { default_cleanup_noexit log_pass } # # Utility function used to list all available pool names. # # NOTE: $KEEP is a variable containing pool names, separated by a newline # character, that must be excluded from the returned list. # function get_all_pools { zpool list -H -o name | grep -Fvx "$KEEP" | grep -v "$NO_POOLS" } function default_cleanup_noexit { typeset pool="" # # Destroying the pool will also destroy any # filesystems it contains. # if is_global_zone; then zfs unmount -a > /dev/null 2>&1 ALL_POOLS=$(get_all_pools) # Here, we loop through the pools we're allowed to # destroy, only destroying them if it's safe to do # so. while [ ! -z ${ALL_POOLS} ] do for pool in ${ALL_POOLS} do if safe_to_destroy_pool $pool ; then destroy_pool $pool fi done ALL_POOLS=$(get_all_pools) done zfs mount -a else typeset fs="" for fs in $(zfs list -H -o name \ | grep "^$ZONE_POOL/$ZONE_CTR[01234]/"); do destroy_dataset "$fs" "-Rf" done # Need cleanup here to avoid garbage dir left. for fs in $(zfs list -H -o name); do [[ $fs == /$ZONE_POOL ]] && continue [[ -d $fs ]] && log_must rm -rf $fs/* done # # Reset the $ZONE_POOL/$ZONE_CTR[01234] file systems property to # the default value # for fs in $(zfs list -H -o name); do if [[ $fs == $ZONE_POOL/$ZONE_CTR[01234] ]]; then log_must zfs set reservation=none $fs log_must zfs set recordsize=128K $fs log_must zfs set mountpoint=/$fs $fs typeset enc=$(get_prop encryption $fs) if [ -z "$enc" ] || [ "$enc" = "off" ]; then log_must zfs set checksum=on $fs fi log_must zfs set compression=off $fs log_must zfs set atime=on $fs log_must zfs set devices=off $fs log_must zfs set exec=on $fs log_must zfs set setuid=on $fs log_must zfs set readonly=off $fs log_must zfs set snapdir=hidden $fs log_must zfs set aclmode=groupmask $fs log_must zfs set aclinherit=secure $fs fi done fi [[ -d $TESTDIR ]] && \ log_must rm -rf $TESTDIR disk1=${DISKS%% *} if is_mpath_device $disk1; then delete_partitions fi rm -f $TEST_BASE_DIR/{err,out} } # # Common function used to cleanup storage pools, file systems # and containers. # function default_container_cleanup { if ! is_global_zone; then reexport_pool fi ismounted $TESTPOOL/$TESTCTR/$TESTFS1 && log_must zfs unmount $TESTPOOL/$TESTCTR/$TESTFS1 destroy_dataset "$TESTPOOL/$TESTCTR/$TESTFS1" "-R" destroy_dataset "$TESTPOOL/$TESTCTR" "-Rf" [[ -e $TESTDIR1 ]] && \ log_must rm -rf $TESTDIR1 default_cleanup } # # Common function used to cleanup snapshot of file system or volume. Default to # delete the file system's snapshot # # $1 snapshot name # function destroy_snapshot { typeset snap=${1:-$TESTPOOL/$TESTFS@$TESTSNAP} if ! snapexists $snap; then log_fail "'$snap' does not exist." fi # # For the sake of the value which come from 'get_prop' is not equal # to the really mountpoint when the snapshot is unmounted. So, firstly # check and make sure this snapshot's been mounted in current system. # typeset mtpt="" if ismounted $snap; then mtpt=$(get_prop mountpoint $snap) fi destroy_dataset "$snap" [[ $mtpt != "" && -d $mtpt ]] && \ log_must rm -rf $mtpt } # # Common function used to cleanup clone. # # $1 clone name # function destroy_clone { typeset clone=${1:-$TESTPOOL/$TESTCLONE} if ! datasetexists $clone; then log_fail "'$clone' does not existed." fi # With the same reason in destroy_snapshot typeset mtpt="" if ismounted $clone; then mtpt=$(get_prop mountpoint $clone) fi destroy_dataset "$clone" [[ $mtpt != "" && -d $mtpt ]] && \ log_must rm -rf $mtpt } # # Common function used to cleanup bookmark of file system or volume. Default # to delete the file system's bookmark. # # $1 bookmark name # function destroy_bookmark { typeset bkmark=${1:-$TESTPOOL/$TESTFS#$TESTBKMARK} if ! bkmarkexists $bkmark; then log_fail "'$bkmarkp' does not existed." fi destroy_dataset "$bkmark" } # Return 0 if a snapshot exists; $? otherwise # # $1 - snapshot name function snapexists { zfs list -H -t snapshot "$1" > /dev/null 2>&1 } # # Return 0 if a bookmark exists; $? otherwise # # $1 - bookmark name # function bkmarkexists { zfs list -H -t bookmark "$1" > /dev/null 2>&1 } # # Return 0 if a hold exists; $? otherwise # # $1 - hold tag # $2 - snapshot name # function holdexists { ! zfs holds "$2" | awk -v t="$1" '$2 ~ t { exit 1 }' } # # Set a property to a certain value on a dataset. # Sets a property of the dataset to the value as passed in. # @param: # $1 dataset who's property is being set # $2 property to set # $3 value to set property to # @return: # 0 if the property could be set. # non-zero otherwise. # @use: ZFS # function dataset_setprop { typeset fn=dataset_setprop if (($# < 3)); then log_note "$fn: Insufficient parameters (need 3, had $#)" return 1 fi typeset output= output=$(zfs set $2=$3 $1 2>&1) typeset rv=$? if ((rv != 0)); then log_note "Setting property on $1 failed." log_note "property $2=$3" log_note "Return Code: $rv" log_note "Output: $output" return $rv fi return 0 } # # Check a numeric assertion # @parameter: $@ the assertion to check # @output: big loud notice if assertion failed # @use: log_fail # function assert { (($@)) || log_fail "$@" } # # Function to format partition size of a disk # Given a disk cxtxdx reduces all partitions # to 0 size # function zero_partitions # { typeset diskname=$1 typeset i if is_freebsd; then gpart destroy -F $diskname elif is_linux; then DSK=$DEV_DSKDIR/$diskname DSK=$(echo $DSK | sed -e "s|//|/|g") log_must parted $DSK -s -- mklabel gpt blockdev --rereadpt $DSK 2>/dev/null block_device_wait else for i in 0 1 3 4 5 6 7 do log_must set_partition $i "" 0mb $diskname done fi return 0 } # # Given a slice, size and disk, this function # formats the slice to the specified size. # Size should be specified with units as per # the `format` command requirements eg. 100mb 3gb # # NOTE: This entire interface is problematic for the Linux parted utility # which requires the end of the partition to be specified. It would be # best to retire this interface and replace it with something more flexible. # At the moment a best effort is made. # # arguments: function set_partition { typeset -i slicenum=$1 typeset start=$2 typeset size=$3 typeset disk=${4#$DEV_DSKDIR/} disk=${disk#$DEV_RDSKDIR/} case "$UNAME" in Linux) if [[ -z $size || -z $disk ]]; then log_fail "The size or disk name is unspecified." fi disk=$DEV_DSKDIR/$disk typeset size_mb=${size%%[mMgG]} size_mb=${size_mb%%[mMgG][bB]} if [[ ${size:1:1} == 'g' ]]; then ((size_mb = size_mb * 1024)) fi # Create GPT partition table when setting slice 0 or # when the device doesn't already contain a GPT label. parted $disk -s -- print 1 >/dev/null typeset ret_val=$? if [[ $slicenum -eq 0 || $ret_val -ne 0 ]]; then if ! parted $disk -s -- mklabel gpt; then log_note "Failed to create GPT partition table on $disk" return 1 fi fi # When no start is given align on the first cylinder. if [[ -z "$start" ]]; then start=1 fi # Determine the cylinder size for the device and using # that calculate the end offset in cylinders. typeset -i cly_size_kb=0 cly_size_kb=$(parted -m $disk -s -- unit cyl print | awk -F '[:k.]' 'NR == 3 {print $4}') ((end = (size_mb * 1024 / cly_size_kb) + start)) parted $disk -s -- \ mkpart part$slicenum ${start}cyl ${end}cyl typeset ret_val=$? if [[ $ret_val -ne 0 ]]; then log_note "Failed to create partition $slicenum on $disk" return 1 fi blockdev --rereadpt $disk 2>/dev/null block_device_wait $disk ;; FreeBSD) if [[ -z $size || -z $disk ]]; then log_fail "The size or disk name is unspecified." fi disk=$DEV_DSKDIR/$disk if [[ $slicenum -eq 0 ]] || ! gpart show $disk >/dev/null 2>&1; then gpart destroy -F $disk >/dev/null 2>&1 if ! gpart create -s GPT $disk; then log_note "Failed to create GPT partition table on $disk" return 1 fi fi typeset index=$((slicenum + 1)) if [[ -n $start ]]; then start="-b $start" fi gpart add -t freebsd-zfs $start -s $size -i $index $disk if [[ $ret_val -ne 0 ]]; then log_note "Failed to create partition $slicenum on $disk" return 1 fi block_device_wait $disk ;; *) if [[ -z $slicenum || -z $size || -z $disk ]]; then log_fail "The slice, size or disk name is unspecified." fi typeset format_file=/var/tmp/format_in.$$ echo "partition" >$format_file echo "$slicenum" >> $format_file echo "" >> $format_file echo "" >> $format_file echo "$start" >> $format_file echo "$size" >> $format_file echo "label" >> $format_file echo "" >> $format_file echo "q" >> $format_file echo "q" >> $format_file format -e -s -d $disk -f $format_file typeset ret_val=$? rm -f $format_file ;; esac if [[ $ret_val -ne 0 ]]; then log_note "Unable to format $disk slice $slicenum to $size" return 1 fi return 0 } # # Delete all partitions on all disks - this is specifically for the use of multipath # devices which currently can only be used in the test suite as raw/un-partitioned # devices (ie a zpool cannot be created on a whole mpath device that has partitions) # function delete_partitions { typeset disk if [[ -z $DISKSARRAY ]]; then DISKSARRAY=$DISKS fi if is_linux; then typeset -i part for disk in $DISKSARRAY; do for (( part = 1; part < MAX_PARTITIONS; part++ )); do typeset partition=${disk}${SLICE_PREFIX}${part} parted $DEV_DSKDIR/$disk -s rm $part > /dev/null 2>&1 if lsblk | grep -qF ${partition}; then log_fail "Partition ${partition} not deleted" else log_note "Partition ${partition} deleted" fi done done elif is_freebsd; then for disk in $DISKSARRAY; do if gpart destroy -F $disk; then log_note "Partitions for ${disk} deleted" else log_fail "Partitions for ${disk} not deleted" fi done fi } # # Get the end cyl of the given slice # function get_endslice # { typeset disk=$1 typeset slice=$2 if [[ -z $disk || -z $slice ]] ; then log_fail "The disk name or slice number is unspecified." fi case "$UNAME" in Linux) endcyl=$(parted -s $DEV_DSKDIR/$disk -- unit cyl print | \ awk "/part${slice}/"' {sub(/cyl/, "", $3); print $3}') ((endcyl = (endcyl + 1))) ;; FreeBSD) disk=${disk#/dev/zvol/} disk=${disk%p*} slice=$((slice + 1)) endcyl=$(gpart show $disk | \ awk -v slice=$slice '$3 == slice { print $1 + $2 }') ;; *) disk=${disk#/dev/dsk/} disk=${disk#/dev/rdsk/} disk=${disk%s*} typeset -i ratio=0 ratio=$(prtvtoc /dev/rdsk/${disk}s2 | \ awk '/sectors\/cylinder/ {print $2}') if ((ratio == 0)); then return fi typeset -i endcyl=$(prtvtoc -h /dev/rdsk/${disk}s2 | awk -v token="$slice" '$1 == token {print $6}') ((endcyl = (endcyl + 1) / ratio)) ;; esac echo $endcyl } # # Given a size,disk and total slice number, this function formats the # disk slices from 0 to the total slice number with the same specified # size. # function partition_disk # { typeset -i i=0 typeset slice_size=$1 typeset disk_name=$2 typeset total_slices=$3 typeset cyl zero_partitions $disk_name while ((i < $total_slices)); do if ! is_linux; then if ((i == 2)); then ((i = i + 1)) continue fi fi log_must set_partition $i "$cyl" $slice_size $disk_name cyl=$(get_endslice $disk_name $i) ((i = i+1)) done } # # This function continues to write to a filenum number of files into dirnum # number of directories until either file_write returns an error or the # maximum number of files per directory have been written. # # Usage: # fill_fs [destdir] [dirnum] [filenum] [bytes] [num_writes] [data] # # Return value: 0 on success # non 0 on error # # Where : # destdir: is the directory where everything is to be created under # dirnum: the maximum number of subdirectories to use, -1 no limit # filenum: the maximum number of files per subdirectory # bytes: number of bytes to write # num_writes: number of types to write out bytes # data: the data that will be written # # E.g. # fill_fs /testdir 20 25 1024 256 0 # # Note: bytes * num_writes equals the size of the testfile # function fill_fs # destdir dirnum filenum bytes num_writes data { typeset destdir=${1:-$TESTDIR} typeset -i dirnum=${2:-50} typeset -i filenum=${3:-50} typeset -i bytes=${4:-8192} typeset -i num_writes=${5:-10240} typeset data=${6:-0} mkdir -p $destdir/{1..$dirnum} for f in $destdir/{1..$dirnum}/$TESTFILE{1..$filenum}; do file_write -o create -f $f -b $bytes -c $num_writes -d $data \ || return done } # Get the specified dataset property in parsable format or fail function get_prop # property dataset { typeset prop=$1 typeset dataset=$2 zfs get -Hpo value "$prop" "$dataset" || log_fail "zfs get $prop $dataset" } # Get the specified pool property in parsable format or fail function get_pool_prop # property pool { typeset prop=$1 typeset pool=$2 zpool get -Hpo value "$prop" "$pool" || log_fail "zpool get $prop $pool" } # Return 0 if a pool exists; $? otherwise # # $1 - pool name function poolexists { typeset pool=$1 if [[ -z $pool ]]; then log_note "No pool name given." return 1 fi zpool get name "$pool" > /dev/null 2>&1 } # Return 0 if all the specified datasets exist; $? otherwise # # $1-n dataset name function datasetexists { if (($# == 0)); then log_note "No dataset name given." return 1 fi zfs get name "$@" > /dev/null 2>&1 } # return 0 if none of the specified datasets exists, otherwise return 1. # # $1-n dataset name function datasetnonexists { if (($# == 0)); then log_note "No dataset name given." return 1 fi while (($# > 0)); do zfs list -H -t filesystem,snapshot,volume $1 > /dev/null 2>&1 \ && return 1 shift done return 0 } # FreeBSD breaks exports(5) at whitespace and doesn't process escapes # Solaris just breaks # # cf. https://github.com/openzfs/zfs/pull/13165#issuecomment-1059845807 # # Linux can have spaces (which are \OOO-escaped), # but can't have backslashes because they're parsed recursively function shares_can_have_whitespace { is_linux } function is_shared_freebsd { typeset fs=$1 pgrep -q mountd && showmount -E | grep -qx "$fs" } function is_shared_illumos { typeset fs=$1 typeset mtpt for mtpt in `share | awk '{print $2}'` ; do if [[ $mtpt == $fs ]] ; then return 0 fi done typeset stat=$(svcs -H -o STA nfs/server:default) if [[ $stat != "ON" ]]; then log_note "Current nfs/server status: $stat" fi return 1 } function is_shared_linux { typeset fs=$1 ! exportfs -s | awk -v fs="${fs//\\/\\\\}" '/^\// && $1 == fs {exit 1}' } # # Given a mountpoint, or a dataset name, determine if it is shared via NFS. # # Returns 0 if shared, 1 otherwise. # function is_shared { typeset fs=$1 typeset mtpt if [[ $fs != "/"* ]] ; then if datasetnonexists "$fs" ; then return 1 else mtpt=$(get_prop mountpoint "$fs") case "$mtpt" in none|legacy|-) return 1 ;; *) fs=$mtpt ;; esac fi fi case "$UNAME" in FreeBSD) is_shared_freebsd "$fs" ;; Linux) is_shared_linux "$fs" ;; *) is_shared_illumos "$fs" ;; esac } function is_exported_illumos { typeset fs=$1 typeset mtpt _ while read -r mtpt _; do [ "$mtpt" = "$fs" ] && return done < /etc/dfs/sharetab return 1 } function is_exported_freebsd { typeset fs=$1 typeset mtpt _ while read -r mtpt _; do [ "$mtpt" = "$fs" ] && return done < /etc/zfs/exports return 1 } function is_exported_linux { typeset fs=$1 typeset mtpt _ while read -r mtpt _; do [ "$(printf "$mtpt")" = "$fs" ] && return done < /etc/exports.d/zfs.exports return 1 } # # Given a mountpoint, or a dataset name, determine if it is exported via # the os-specific NFS exports file. # # Returns 0 if exported, 1 otherwise. # function is_exported { typeset fs=$1 typeset mtpt if [[ $fs != "/"* ]] ; then if datasetnonexists "$fs" ; then return 1 else mtpt=$(get_prop mountpoint "$fs") case $mtpt in none|legacy|-) return 1 ;; *) fs=$mtpt ;; esac fi fi case "$UNAME" in FreeBSD) is_exported_freebsd "$fs" ;; Linux) is_exported_linux "$fs" ;; *) is_exported_illumos "$fs" ;; esac } # # Given a dataset name determine if it is shared via SMB. # # Returns 0 if shared, 1 otherwise. # function is_shared_smb { typeset fs=$1 datasetexists "$fs" || return if is_linux; then net usershare list | grep -xFq "${fs//[-\/]/_}" else log_note "SMB on $UNAME currently unsupported by the test framework" return 1 fi } # # Given a mountpoint, determine if it is not shared via NFS. # # Returns 0 if not shared, 1 otherwise. # function not_shared { ! is_shared $1 } # # Given a dataset determine if it is not shared via SMB. # # Returns 0 if not shared, 1 otherwise. # function not_shared_smb { ! is_shared_smb $1 } # # Helper function to unshare a mountpoint. # function unshare_fs #fs { typeset fs=$1 if is_shared $fs || is_shared_smb $fs; then log_must zfs unshare $fs fi } # # Helper function to share a NFS mountpoint. # function share_nfs #fs { typeset fs=$1 is_shared "$fs" && return case "$UNAME" in Linux) log_must exportfs "*:$fs" ;; FreeBSD) typeset mountd read -r mountd < /var/run/mountd.pid log_must eval "printf '%s\t\n' \"$fs\" >> /etc/zfs/exports" log_must kill -s HUP "$mountd" ;; *) log_must share -F nfs "$fs" ;; esac return 0 } # # Helper function to unshare a NFS mountpoint. # function unshare_nfs #fs { typeset fs=$1 ! is_shared "$fs" && return case "$UNAME" in Linux) log_must exportfs -u "*:$fs" ;; FreeBSD) typeset mountd read -r mountd < /var/run/mountd.pid awk -v fs="${fs//\\/\\\\}" '$1 != fs' /etc/zfs/exports > /etc/zfs/exports.$$ log_must mv /etc/zfs/exports.$$ /etc/zfs/exports log_must kill -s HUP "$mountd" ;; *) log_must unshare -F nfs $fs ;; esac return 0 } # # Helper function to show NFS shares. # function showshares_nfs { case "$UNAME" in Linux) exportfs -v ;; FreeBSD) showmount ;; *) share -F nfs ;; esac } function check_nfs { case "$UNAME" in Linux) exportfs -s ;; FreeBSD) showmount -e ;; *) log_unsupported "Unknown platform" ;; esac || log_unsupported "The NFS utilities are not installed" } # # Check NFS server status and trigger it online. # function setup_nfs_server { # Cannot share directory in non-global zone. # if ! is_global_zone; then log_note "Cannot trigger NFS server by sharing in LZ." return fi if is_linux; then # # Re-synchronize /var/lib/nfs/etab with /etc/exports and # /etc/exports.d./* to provide a clean test environment. # log_must exportfs -r log_note "NFS server must be started prior to running ZTS." return elif is_freebsd; then log_must kill -s HUP $(/dev/null) [ $cur_zone = "global" ] fi } # # Verify whether test is permitted to run from # global zone, local zone, or both # # $1 zone limit, could be "global", "local", or "both"(no limit) # # Return 0 if permitted, otherwise exit with log_unsupported # function verify_runnable # zone limit { typeset limit=$1 [[ -z $limit ]] && return 0 if is_global_zone ; then case $limit in global|both) ;; local) log_unsupported "Test is unable to run from "\ "global zone." ;; *) log_note "Warning: unknown limit $limit - " \ "use both." ;; esac else case $limit in local|both) ;; global) log_unsupported "Test is unable to run from "\ "local zone." ;; *) log_note "Warning: unknown limit $limit - " \ "use both." ;; esac reexport_pool fi return 0 } # Return 0 if create successfully or the pool exists; $? otherwise # Note: In local zones, this function should return 0 silently. # # $1 - pool name # $2-n - [keyword] devs_list function create_pool #pool devs_list { typeset pool=${1%%/*} shift if [[ -z $pool ]]; then log_note "Missing pool name." return 1 fi if poolexists $pool ; then destroy_pool $pool fi if is_global_zone ; then [[ -d /$pool ]] && rm -rf /$pool log_must zpool create -f $pool $@ fi return 0 } # Return 0 if destroy successfully or the pool exists; $? otherwise # Note: In local zones, this function should return 0 silently. # # $1 - pool name # Destroy pool with the given parameters. function destroy_pool #pool { typeset pool=${1%%/*} typeset mtpt if [[ -z $pool ]]; then log_note "No pool name given." return 1 fi if is_global_zone ; then if poolexists "$pool" ; then mtpt=$(get_prop mountpoint "$pool") # At times, syseventd/udev activity can cause attempts # to destroy a pool to fail with EBUSY. We retry a few # times allowing failures before requiring the destroy # to succeed. log_must_busy zpool destroy -f $pool [[ -d $mtpt ]] && \ log_must rm -rf $mtpt else log_note "Pool does not exist. ($pool)" return 1 fi fi return 0 } # Return 0 if created successfully; $? otherwise # # $1 - dataset name # $2-n - dataset options function create_dataset #dataset dataset_options { typeset dataset=$1 shift if [[ -z $dataset ]]; then log_note "Missing dataset name." return 1 fi if datasetexists $dataset ; then destroy_dataset $dataset fi log_must zfs create $@ $dataset return 0 } # Return 0 if destroy successfully or the dataset exists; $? otherwise # Note: In local zones, this function should return 0 silently. # # $1 - dataset name # $2 - custom arguments for zfs destroy # Destroy dataset with the given parameters. function destroy_dataset # dataset [args] { typeset dataset=$1 typeset mtpt typeset args=${2:-""} if [[ -z $dataset ]]; then log_note "No dataset name given." return 1 fi if is_global_zone ; then if datasetexists "$dataset" ; then mtpt=$(get_prop mountpoint "$dataset") log_must_busy zfs destroy $args $dataset [ -d $mtpt ] && log_must rm -rf $mtpt else log_note "Dataset does not exist. ($dataset)" return 1 fi fi return 0 } # # Reexport TESTPOOL & TESTPOOL(1-4) # function reexport_pool { typeset -i cntctr=5 typeset -i i=0 while ((i < cntctr)); do if ((i == 0)); then TESTPOOL=$ZONE_POOL/$ZONE_CTR$i if ! ismounted $TESTPOOL; then log_must zfs mount $TESTPOOL fi else eval TESTPOOL$i=$ZONE_POOL/$ZONE_CTR$i if eval ! ismounted \$TESTPOOL$i; then log_must eval zfs mount \$TESTPOOL$i fi fi ((i += 1)) done } # # Verify a given disk or pool state # # Return 0 is pool/disk matches expected state, 1 otherwise # function check_state # pool disk state{online,offline,degraded} { typeset pool=$1 typeset disk=${2#$DEV_DSKDIR/} typeset state=$3 [[ -z $pool ]] || [[ -z $state ]] \ && log_fail "Arguments invalid or missing" if [[ -z $disk ]]; then #check pool state only zpool get -H -o value health $pool | grep -qi "$state" else zpool status -v $pool | grep "$disk" | grep -qi "$state" fi } # # Get the mountpoint of snapshot # For the snapshot use /.zfs/snapshot/ # as its mountpoint # function snapshot_mountpoint { typeset dataset=${1:-$TESTPOOL/$TESTFS@$TESTSNAP} if [[ $dataset != *@* ]]; then log_fail "Error name of snapshot '$dataset'." fi typeset fs=${dataset%@*} typeset snap=${dataset#*@} if [[ -z $fs || -z $snap ]]; then log_fail "Error name of snapshot '$dataset'." fi echo $(get_prop mountpoint $fs)/.zfs/snapshot/$snap } # # Given a device and 'ashift' value verify it's correctly set on every label # function verify_ashift # device ashift { typeset device="$1" typeset ashift="$2" zdb -e -lll $device | awk -v ashift=$ashift ' /ashift: / { if (ashift != $2) exit 1; else count++; } END { exit (count != 4); }' } # # Given a pool and file system, this function will verify the file system # using the zdb internal tool. Note that the pool is exported and imported # to ensure it has consistent state. # function verify_filesys # pool filesystem dir { typeset pool="$1" typeset filesys="$2" typeset zdbout="/tmp/zdbout.$$" shift shift typeset dirs=$@ typeset search_path="" log_note "Calling zdb to verify filesystem '$filesys'" zfs unmount -a > /dev/null 2>&1 log_must zpool export $pool if [[ -n $dirs ]] ; then for dir in $dirs ; do search_path="$search_path -d $dir" done fi log_must zpool import $search_path $pool if ! zdb -cudi $filesys > $zdbout 2>&1; then log_note "Output: zdb -cudi $filesys" cat $zdbout rm -f $zdbout log_fail "zdb detected errors with: '$filesys'" fi log_must zfs mount -a log_must rm -rf $zdbout } # # Given a pool issue a scrub and verify that no checksum errors are reported. # function verify_pool { typeset pool=${1:-$TESTPOOL} log_must zpool scrub $pool log_must wait_scrubbed $pool typeset -i cksum=$(zpool status $pool | awk ' !NF { isvdev = 0 } isvdev { errors += $NF } /CKSUM$/ { isvdev = 1 } END { print errors } ') if [[ $cksum != 0 ]]; then log_must zpool status -v log_fail "Unexpected CKSUM errors found on $pool ($cksum)" fi } # # Given a pool, and this function list all disks in the pool # function get_disklist # pool { echo $(zpool iostat -v $1 | awk '(NR > 4) {print $1}' | \ grep -vEe '^-----' -e "^(mirror|raidz[1-3]|draid[1-3]|spare|log|cache|special|dedup)|\-[0-9]$") } # # Given a pool, and this function list all disks in the pool with their full # path (like "/dev/sda" instead of "sda"). # function get_disklist_fullpath # pool { get_disklist "-P $1" } # /** # This function kills a given list of processes after a time period. We use # this in the stress tests instead of STF_TIMEOUT so that we can have processes # run for a fixed amount of time, yet still pass. Tests that hit STF_TIMEOUT # would be listed as FAIL, which we don't want : we're happy with stress tests # running for a certain amount of time, then finishing. # # @param $1 the time in seconds after which we should terminate these processes # @param $2..$n the processes we wish to terminate. # */ function stress_timeout { typeset -i TIMEOUT=$1 shift typeset cpids="$@" log_note "Waiting for child processes($cpids). " \ "It could last dozens of minutes, please be patient ..." log_must sleep $TIMEOUT log_note "Killing child processes after ${TIMEOUT} stress timeout." typeset pid for pid in $cpids; do ps -p $pid > /dev/null 2>&1 && log_must kill -USR1 $pid done } # # Verify a given hotspare disk is inuse or avail # # Return 0 is pool/disk matches expected state, 1 otherwise # function check_hotspare_state # pool disk state{inuse,avail} { typeset pool=$1 typeset disk=${2#$DEV_DSKDIR/} typeset state=$3 cur_state=$(get_device_state $pool $disk "spares") [ $state = $cur_state ] } # # Wait until a hotspare transitions to a given state or times out. # # Return 0 when pool/disk matches expected state, 1 on timeout. # function wait_hotspare_state # pool disk state timeout { typeset pool=$1 typeset disk=${2#*$DEV_DSKDIR/} typeset state=$3 typeset timeout=${4:-60} typeset -i i=0 while [[ $i -lt $timeout ]]; do if check_hotspare_state $pool $disk $state; then return 0 fi i=$((i+1)) sleep 1 done return 1 } # # Verify a given vdev disk is inuse or avail # # Return 0 is pool/disk matches expected state, 1 otherwise # -function check_vdev_state # pool disk state{online,offline,unavail} +function check_vdev_state # pool disk state{online,offline,unavail,removed} { typeset pool=$1 typeset disk=${2#*$DEV_DSKDIR/} typeset state=$3 cur_state=$(get_device_state $pool $disk) [ $state = $cur_state ] } # # Wait until a vdev transitions to a given state or times out. # # Return 0 when pool/disk matches expected state, 1 on timeout. # function wait_vdev_state # pool disk state timeout { typeset pool=$1 typeset disk=${2#*$DEV_DSKDIR/} typeset state=$3 typeset timeout=${4:-60} typeset -i i=0 while [[ $i -lt $timeout ]]; do if check_vdev_state $pool $disk $state; then return 0 fi i=$((i+1)) sleep 1 done return 1 } # # Check the output of 'zpool status -v ', # and to see if the content of contain the specified. # # Return 0 is contain, 1 otherwise # function check_pool_status # pool token keyword { typeset pool=$1 typeset token=$2 typeset keyword=$3 typeset verbose=${4:-false} scan=$(zpool status -v "$pool" 2>/dev/null | awk -v token="$token:" '$1==token') if [[ $verbose == true ]]; then log_note $scan fi echo $scan | grep -qi "$keyword" } # # The following functions are instance of check_pool_status() # is_pool_resilvering - to check if the pool resilver is in progress # is_pool_resilvered - to check if the pool resilver is completed # is_pool_scrubbing - to check if the pool scrub is in progress # is_pool_scrubbed - to check if the pool scrub is completed # is_pool_scrub_stopped - to check if the pool scrub is stopped # is_pool_scrub_paused - to check if the pool scrub has paused # is_pool_removing - to check if the pool removing is a vdev # is_pool_removed - to check if the pool remove is completed # is_pool_discarding - to check if the pool checkpoint is being discarded # function is_pool_resilvering #pool { check_pool_status "$1" "scan" \ "resilver[ ()0-9A-Za-z:_-]* in progress since" $2 } function is_pool_resilvered #pool { check_pool_status "$1" "scan" "resilvered " $2 } function is_pool_scrubbing #pool { check_pool_status "$1" "scan" "scrub in progress since " $2 } function is_pool_scrubbed #pool { check_pool_status "$1" "scan" "scrub repaired" $2 } function is_pool_scrub_stopped #pool { check_pool_status "$1" "scan" "scrub canceled" $2 } function is_pool_scrub_paused #pool { check_pool_status "$1" "scan" "scrub paused since " $2 } function is_pool_removing #pool { check_pool_status "$1" "remove" "in progress since " } function is_pool_removed #pool { check_pool_status "$1" "remove" "completed on" } function is_pool_discarding #pool { check_pool_status "$1" "checkpoint" "discarding" } function wait_for_degraded { typeset pool=$1 typeset timeout=${2:-30} typeset t0=$SECONDS while :; do [[ $(get_pool_prop health $pool) == "DEGRADED" ]] && break log_note "$pool is not yet degraded." sleep 1 if ((SECONDS - t0 > $timeout)); then log_note "$pool not degraded after $timeout seconds." return 1 fi done return 0 } # # Use create_pool()/destroy_pool() to clean up the information in # in the given disk to avoid slice overlapping. # function cleanup_devices #vdevs { typeset pool="foopool$$" for vdev in $@; do zero_partitions $vdev done poolexists $pool && destroy_pool $pool create_pool $pool $@ destroy_pool $pool return 0 } #/** # A function to find and locate free disks on a system or from given # disks as the parameter. It works by locating disks that are in use # as swap devices and dump devices, and also disks listed in /etc/vfstab # # $@ given disks to find which are free, default is all disks in # the test system # # @return a string containing the list of available disks #*/ function find_disks { # Trust provided list, no attempt is made to locate unused devices. if is_linux || is_freebsd; then echo "$@" return fi sfi=/tmp/swaplist.$$ dmpi=/tmp/dumpdev.$$ max_finddisksnum=${MAX_FINDDISKSNUM:-6} swap -l > $sfi dumpadm > $dmpi 2>/dev/null disks=${@:-$(echo "" | format -e 2>/dev/null | awk ' BEGIN { FS="."; } /^Specify disk/{ searchdisks=0; } { if (searchdisks && $2 !~ "^$"){ split($2,arr," "); print arr[1]; } } /^AVAILABLE DISK SELECTIONS:/{ searchdisks=1; } ')} unused="" for disk in $disks; do # Check for mounted grep -q "${disk}[sp]" /etc/mnttab && continue # Check for swap grep -q "${disk}[sp]" $sfi && continue # check for dump device grep -q "${disk}[sp]" $dmpi && continue # check to see if this disk hasn't been explicitly excluded # by a user-set environment variable echo "${ZFS_HOST_DEVICES_IGNORE}" | grep -q "${disk}" && continue unused_candidates="$unused_candidates $disk" done rm $sfi $dmpi # now just check to see if those disks do actually exist # by looking for a device pointing to the first slice in # each case. limit the number to max_finddisksnum count=0 for disk in $unused_candidates; do if is_disk_device $DEV_DSKDIR/${disk}s0 && \ [ $count -lt $max_finddisksnum ]; then unused="$unused $disk" # do not impose limit if $@ is provided [[ -z $@ ]] && ((count = count + 1)) fi done # finally, return our disk list echo $unused } function add_user_freebsd # { typeset group=$1 typeset user=$2 typeset basedir=$3 # Check to see if the user exists. if id $user > /dev/null 2>&1; then return 0 fi # Assign 1000 as the base uid typeset -i uid=1000 while true; do pw useradd -u $uid -g $group -d $basedir/$user -m -n $user case $? in 0) break ;; # The uid is not unique 65) ((uid += 1)) ;; *) return 1 ;; esac if [[ $uid == 65000 ]]; then log_fail "No user id available under 65000 for $user" fi done # Silence MOTD touch $basedir/$user/.hushlogin return 0 } # # Delete the specified user. # # $1 login name # function del_user_freebsd # { typeset user=$1 if id $user > /dev/null 2>&1; then log_must pw userdel $user fi return 0 } # # Select valid gid and create specified group. # # $1 group name # function add_group_freebsd # { typeset group=$1 # See if the group already exists. if pw groupshow $group >/dev/null 2>&1; then return 0 fi # Assign 1000 as the base gid typeset -i gid=1000 while true; do pw groupadd -g $gid -n $group > /dev/null 2>&1 case $? in 0) return 0 ;; # The gid is not unique 65) ((gid += 1)) ;; *) return 1 ;; esac if [[ $gid == 65000 ]]; then log_fail "No user id available under 65000 for $group" fi done } # # Delete the specified group. # # $1 group name # function del_group_freebsd # { typeset group=$1 pw groupdel -n $group > /dev/null 2>&1 case $? in # Group does not exist, or was deleted successfully. 0|6|65) return 0 ;; # Name already exists as a group name 9) log_must pw groupdel $group ;; *) return 1 ;; esac return 0 } function add_user_illumos # { typeset group=$1 typeset user=$2 typeset basedir=$3 log_must useradd -g $group -d $basedir/$user -m $user return 0 } function del_user_illumos # { typeset user=$1 if id $user > /dev/null 2>&1; then log_must_retry "currently used" 6 userdel $user fi return 0 } function add_group_illumos # { typeset group=$1 typeset -i gid=100 while true; do groupadd -g $gid $group > /dev/null 2>&1 case $? in 0) return 0 ;; # The gid is not unique 4) ((gid += 1)) ;; *) return 1 ;; esac done } function del_group_illumos # { typeset group=$1 groupmod -n $grp $grp > /dev/null 2>&1 case $? in # Group does not exist. 6) return 0 ;; # Name already exists as a group name 9) log_must groupdel $grp ;; *) return 1 ;; esac } function add_user_linux # { typeset group=$1 typeset user=$2 typeset basedir=$3 log_must useradd -g $group -d $basedir/$user -m $user # Add new users to the same group and the command line utils. # This allows them to be run out of the original users home # directory as long as it permissioned to be group readable. cmd_group=$(stat --format="%G" $(command -v zfs)) log_must usermod -a -G $cmd_group $user return 0 } function del_user_linux # { typeset user=$1 if id $user > /dev/null 2>&1; then log_must_retry "currently used" 6 userdel $user fi } function add_group_linux # { typeset group=$1 # Assign 100 as the base gid, a larger value is selected for # Linux because for many distributions 1000 and under are reserved. while true; do groupadd $group > /dev/null 2>&1 case $? in 0) return 0 ;; *) return 1 ;; esac done } function del_group_linux # { typeset group=$1 getent group $group > /dev/null 2>&1 case $? in # Group does not exist. 2) return 0 ;; # Name already exists as a group name 0) log_must groupdel $group ;; *) return 1 ;; esac return 0 } # # Add specified user to specified group # # $1 group name # $2 user name # $3 base of the homedir (optional) # function add_user # { typeset group=$1 typeset user=$2 typeset basedir=${3:-"/var/tmp"} if ((${#group} == 0 || ${#user} == 0)); then log_fail "group name or user name are not defined." fi case "$UNAME" in FreeBSD) add_user_freebsd "$group" "$user" "$basedir" ;; Linux) add_user_linux "$group" "$user" "$basedir" ;; *) add_user_illumos "$group" "$user" "$basedir" ;; esac return 0 } # # Delete the specified user. # # $1 login name # $2 base of the homedir (optional) # function del_user # { typeset user=$1 typeset basedir=${2:-"/var/tmp"} if ((${#user} == 0)); then log_fail "login name is necessary." fi case "$UNAME" in FreeBSD) del_user_freebsd "$user" ;; Linux) del_user_linux "$user" ;; *) del_user_illumos "$user" ;; esac [[ -d $basedir/$user ]] && rm -fr $basedir/$user return 0 } # # Select valid gid and create specified group. # # $1 group name # function add_group # { typeset group=$1 if ((${#group} == 0)); then log_fail "group name is necessary." fi case "$UNAME" in FreeBSD) add_group_freebsd "$group" ;; Linux) add_group_linux "$group" ;; *) add_group_illumos "$group" ;; esac return 0 } # # Delete the specified group. # # $1 group name # function del_group # { typeset group=$1 if ((${#group} == 0)); then log_fail "group name is necessary." fi case "$UNAME" in FreeBSD) del_group_freebsd "$group" ;; Linux) del_group_linux "$group" ;; *) del_group_illumos "$group" ;; esac return 0 } # # This function will return true if it's safe to destroy the pool passed # as argument 1. It checks for pools based on zvols and files, and also # files contained in a pool that may have a different mountpoint. # function safe_to_destroy_pool { # $1 the pool name typeset pool="" typeset DONT_DESTROY="" # We check that by deleting the $1 pool, we're not # going to pull the rug out from other pools. Do this # by looking at all other pools, ensuring that they # aren't built from files or zvols contained in this pool. for pool in $(zpool list -H -o name) do ALTMOUNTPOOL="" # this is a list of the top-level directories in each of the # files that make up the path to the files the pool is based on FILEPOOL=$(zpool status -v $pool | awk -v pool="/$1/" '$0 ~ pool {print $1}') # this is a list of the zvols that make up the pool ZVOLPOOL=$(zpool status -v $pool | awk -v zvols="$ZVOL_DEVDIR/$1$" '$0 ~ zvols {print $1}') # also want to determine if it's a file-based pool using an # alternate mountpoint... POOL_FILE_DIRS=$(zpool status -v $pool | \ awk '/\// {print $1}' | \ awk -F/ '!/dev/ {print $2}') for pooldir in $POOL_FILE_DIRS do OUTPUT=$(zfs list -H -r -o mountpoint $1 | \ awk -v pd="${pooldir}$" '$0 ~ pd {print $1}') ALTMOUNTPOOL="${ALTMOUNTPOOL}${OUTPUT}" done if [ ! -z "$ZVOLPOOL" ] then DONT_DESTROY="true" log_note "Pool $pool is built from $ZVOLPOOL on $1" fi if [ ! -z "$FILEPOOL" ] then DONT_DESTROY="true" log_note "Pool $pool is built from $FILEPOOL on $1" fi if [ ! -z "$ALTMOUNTPOOL" ] then DONT_DESTROY="true" log_note "Pool $pool is built from $ALTMOUNTPOOL on $1" fi done if [ -z "${DONT_DESTROY}" ] then return 0 else log_note "Warning: it is not safe to destroy $1!" return 1 fi } # # Verify zfs operation with -p option work as expected # $1 operation, value could be create, clone or rename # $2 dataset type, value could be fs or vol # $3 dataset name # $4 new dataset name # function verify_opt_p_ops { typeset ops=$1 typeset datatype=$2 typeset dataset=$3 typeset newdataset=$4 if [[ $datatype != "fs" && $datatype != "vol" ]]; then log_fail "$datatype is not supported." fi # check parameters accordingly case $ops in create) newdataset=$dataset dataset="" if [[ $datatype == "vol" ]]; then ops="create -V $VOLSIZE" fi ;; clone) if [[ -z $newdataset ]]; then log_fail "newdataset should not be empty" \ "when ops is $ops." fi log_must datasetexists $dataset log_must snapexists $dataset ;; rename) if [[ -z $newdataset ]]; then log_fail "newdataset should not be empty" \ "when ops is $ops." fi log_must datasetexists $dataset ;; *) log_fail "$ops is not supported." ;; esac # make sure the upper level filesystem does not exist destroy_dataset "${newdataset%/*}" "-rRf" # without -p option, operation will fail log_mustnot zfs $ops $dataset $newdataset log_mustnot datasetexists $newdataset ${newdataset%/*} # with -p option, operation should succeed log_must zfs $ops -p $dataset $newdataset block_device_wait if ! datasetexists $newdataset ; then log_fail "-p option does not work for $ops" fi # when $ops is create or clone, redo the operation still return zero if [[ $ops != "rename" ]]; then log_must zfs $ops -p $dataset $newdataset fi return 0 } # # Get configuration of pool # $1 pool name # $2 config name # function get_config { typeset pool=$1 typeset config=$2 if ! poolexists "$pool" ; then return 1 fi if [ "$(get_pool_prop cachefile "$pool")" = "none" ]; then zdb -e $pool else zdb -C $pool fi | awk -F: -v cfg="$config:" '$0 ~ cfg {sub(/^'\''/, $2); sub(/'\''$/, $2); print $2}' } # # Privated function. Random select one of items from arguments. # # $1 count # $2-n string # function _random_get { typeset cnt=$1 shift typeset str="$@" typeset -i ind ((ind = RANDOM % cnt + 1)) echo "$str" | cut -f $ind -d ' ' } # # Random select one of item from arguments which include NONE string # function random_get_with_non { typeset -i cnt=$# ((cnt =+ 1)) _random_get "$cnt" "$@" } # # Random select one of item from arguments which doesn't include NONE string # function random_get { _random_get "$#" "$@" } # # The function will generate a dataset name with specific length # $1, the length of the name # $2, the base string to construct the name # function gen_dataset_name { typeset -i len=$1 typeset basestr="$2" typeset -i baselen=${#basestr} typeset -i iter=0 typeset l_name="" if ((len % baselen == 0)); then ((iter = len / baselen)) else ((iter = len / baselen + 1)) fi while ((iter > 0)); do l_name="${l_name}$basestr" ((iter -= 1)) done echo $l_name } # # Get cksum tuple of dataset # $1 dataset name # # sample zdb output: # Dataset data/test [ZPL], ID 355, cr_txg 2413856, 31.0K, 7 objects, rootbp # DVA[0]=<0:803046400:200> DVA[1]=<0:81199000:200> [L0 DMU objset] fletcher4 # lzjb LE contiguous unique double size=800L/200P birth=2413856L/2413856P # fill=7 cksum=11ce125712:643a9c18ee2:125e25238fca0:254a3f74b59744 function datasetcksum { typeset cksum sync sync_all_pools zdb -vvv $1 | awk -F= -v ds="^Dataset $1 "'\\[' '$0 ~ ds && /cksum/ {print $7}' } # # Get the given disk/slice state from the specific field of the pool # function get_device_state #pool disk field("", "spares","logs") { typeset pool=$1 typeset disk=${2#$DEV_DSKDIR/} typeset field=${3:-$pool} zpool status -v "$pool" 2>/dev/null | \ awk -v device=$disk -v pool=$pool -v field=$field \ 'BEGIN {startconfig=0; startfield=0; } /config:/ {startconfig=1} (startconfig==1) && ($1==field) {startfield=1; next;} (startfield==1) && ($1==device) {print $2; exit;} (startfield==1) && ($1==field || $1 ~ "^spares$" || $1 ~ "^logs$") {startfield=0}' } # # get the root filesystem name if it's zfsroot system. # # return: root filesystem name function get_rootfs { typeset rootfs="" if is_freebsd; then rootfs=$(mount -p | awk '$2 == "/" && $3 == "zfs" {print $1}') elif ! is_linux; then rootfs=$(awk '$2 == "/" && $3 == "zfs" {print $1}' \ /etc/mnttab) fi if [[ -z "$rootfs" ]]; then log_fail "Can not get rootfs" fi if datasetexists $rootfs; then echo $rootfs else log_fail "This is not a zfsroot system." fi } # # get the rootfs's pool name # return: # rootpool name # function get_rootpool { typeset rootfs=$(get_rootfs) echo ${rootfs%%/*} } # # To verify if the require numbers of disks is given # function verify_disk_count { typeset -i min=${2:-1} typeset -i count=$(echo "$1" | wc -w) if ((count < min)); then log_untested "A minimum of $min disks is required to run." \ " You specified $count disk(s)" fi } function ds_is_volume { typeset type=$(get_prop type $1) [ $type = "volume" ] } function ds_is_filesystem { typeset type=$(get_prop type $1) [ $type = "filesystem" ] } # # Check if Trusted Extensions are installed and enabled # function is_te_enabled { svcs -H -o state labeld 2>/dev/null | grep -q "enabled" } # Return the number of CPUs (cross-platform) function get_num_cpus { if is_linux ; then grep -c '^processor' /proc/cpuinfo elif is_freebsd; then sysctl -n kern.smp.cpus else psrinfo | wc -l fi } # Utility function to determine if a system has multiple cpus. function is_mp { [[ $(get_num_cpus) -gt 1 ]] } function get_cpu_freq { if is_linux; then lscpu | awk '/CPU MHz/ { print $3 }' elif is_freebsd; then sysctl -n hw.clockrate else psrinfo -v 0 | awk '/processor operates at/ {print $6}' fi } # Run the given command as the user provided. function user_run { typeset user=$1 shift log_note "user: $user" log_note "cmd: $*" typeset out=$TEST_BASE_DIR/out typeset err=$TEST_BASE_DIR/err sudo -Eu $user env PATH="$PATH" ksh <<<"$*" >$out 2>$err typeset res=$? log_note "out: $(<$out)" log_note "err: $(<$err)" return $res } # # Check if the pool contains the specified vdevs # # $1 pool # $2..n ... # # Return 0 if the vdevs are contained in the pool, 1 if any of the specified # vdevs is not in the pool, and 2 if pool name is missing. # function vdevs_in_pool { typeset pool=$1 typeset vdev if [[ -z $pool ]]; then log_note "Missing pool name." return 2 fi shift # We could use 'zpool list' to only get the vdevs of the pool but we # can't reference a mirror/raidz vdev using its ID (i.e mirror-0), # therefore we use the 'zpool status' output. typeset tmpfile=$(mktemp) zpool status -v "$pool" | grep -A 1000 "config:" >$tmpfile for vdev in "$@"; do grep -wq ${vdev##*/} $tmpfile || return 1 done rm -f $tmpfile return 0 } function get_max { typeset -l i max=$1 shift for i in "$@"; do max=$((max > i ? max : i)) done echo $max } # Write data that can be compressed into a directory function write_compressible { typeset dir=$1 typeset megs=$2 typeset nfiles=${3:-1} typeset bs=${4:-1024k} typeset fname=${5:-file} [[ -d $dir ]] || log_fail "No directory: $dir" # Under Linux fio is not currently used since its behavior can # differ significantly across versions. This includes missing # command line options and cases where the --buffer_compress_* # options fail to behave as expected. if is_linux; then typeset file_bytes=$(to_bytes $megs) typeset bs_bytes=4096 typeset blocks=$(($file_bytes / $bs_bytes)) for (( i = 0; i < $nfiles; i++ )); do truncate -s $file_bytes $dir/$fname.$i # Write every third block to get 66% compression. for (( j = 0; j < $blocks; j += 3 )); do dd if=/dev/urandom of=$dir/$fname.$i \ seek=$j bs=$bs_bytes count=1 \ conv=notrunc >/dev/null 2>&1 done done else command -v fio > /dev/null || log_unsupported "fio missing" log_must eval fio \ --name=job \ --fallocate=0 \ --minimal \ --randrepeat=0 \ --buffer_compress_percentage=66 \ --buffer_compress_chunk=4096 \ --directory="$dir" \ --numjobs="$nfiles" \ --nrfiles="$nfiles" \ --rw=write \ --bs="$bs" \ --filesize="$megs" \ "--filename_format='$fname.\$jobnum' >/dev/null" fi } function get_objnum { typeset pathname=$1 typeset objnum [[ -e $pathname ]] || log_fail "No such file or directory: $pathname" if is_freebsd; then objnum=$(stat -f "%i" $pathname) else objnum=$(stat -c %i $pathname) fi echo $objnum } # # Sync data to the pool # # $1 pool name # $2 boolean to force uberblock (and config including zpool cache file) update # function sync_pool #pool { typeset pool=${1:-$TESTPOOL} typeset force=${2:-false} if [[ $force == true ]]; then log_must zpool sync -f $pool else log_must zpool sync $pool fi return 0 } # # Sync all pools # # $1 boolean to force uberblock (and config including zpool cache file) update # function sync_all_pools # { typeset force=${1:-false} if [[ $force == true ]]; then log_must zpool sync -f else log_must zpool sync fi return 0 } # # Wait for zpool 'freeing' property drops to zero. # # $1 pool name # function wait_freeing #pool { typeset pool=${1:-$TESTPOOL} while true; do [[ "0" == "$(zpool list -Ho freeing $pool)" ]] && break log_must sleep 1 done } # # Wait for every device replace operation to complete # # $1 pool name # function wait_replacing #pool { typeset pool=${1:-$TESTPOOL} while zpool status $pool | grep -qE 'replacing-[0-9]+'; do log_must sleep 1 done } # Wait for a pool to be scrubbed # # $1 pool name # $2 timeout # function wait_scrubbed #pool timeout { typeset timeout=${2:-300} typeset pool=${1:-$TESTPOOL} for (( timer = 0; timer < $timeout; timer++ )); do is_pool_scrubbed $pool && break; sleep 1; done } # Backup the zed.rc in our test directory so that we can edit it for our test. # # Returns: Backup file name. You will need to pass this to zed_rc_restore(). function zed_rc_backup { zedrc_backup="$(mktemp)" cp $ZEDLET_DIR/zed.rc $zedrc_backup echo $zedrc_backup } function zed_rc_restore { mv $1 $ZEDLET_DIR/zed.rc } # # Setup custom environment for the ZED. # # $@ Optional list of zedlets to run under zed. function zed_setup { if ! is_linux; then log_unsupported "No zed on $UNAME" fi if [[ ! -d $ZEDLET_DIR ]]; then log_must mkdir $ZEDLET_DIR fi if [[ ! -e $VDEVID_CONF ]]; then log_must touch $VDEVID_CONF fi if [[ -e $VDEVID_CONF_ETC ]]; then log_fail "Must not have $VDEVID_CONF_ETC file present on system" fi EXTRA_ZEDLETS=$@ # Create a symlink for /etc/zfs/vdev_id.conf file. log_must ln -s $VDEVID_CONF $VDEVID_CONF_ETC # Setup minimal ZED configuration. Individual test cases should # add additional ZEDLETs as needed for their specific test. log_must cp ${ZEDLET_ETC_DIR}/zed.rc $ZEDLET_DIR log_must cp ${ZEDLET_ETC_DIR}/zed-functions.sh $ZEDLET_DIR # Scripts must only be user writable. if [[ -n "$EXTRA_ZEDLETS" ]] ; then saved_umask=$(umask) log_must umask 0022 for i in $EXTRA_ZEDLETS ; do log_must cp ${ZEDLET_LIBEXEC_DIR}/$i $ZEDLET_DIR done log_must umask $saved_umask fi # Customize the zed.rc file to enable the full debug log. log_must sed -i '/\#ZED_DEBUG_LOG=.*/d' $ZEDLET_DIR/zed.rc echo "ZED_DEBUG_LOG=$ZED_DEBUG_LOG" >>$ZEDLET_DIR/zed.rc } # # Cleanup custom ZED environment. # # $@ Optional list of zedlets to remove from our test zed.d directory. function zed_cleanup { if ! is_linux; then return fi for extra_zedlet; do log_must rm -f ${ZEDLET_DIR}/$extra_zedlet done log_must rm -fd ${ZEDLET_DIR}/zed.rc ${ZEDLET_DIR}/zed-functions.sh ${ZEDLET_DIR}/all-syslog.sh ${ZEDLET_DIR}/all-debug.sh ${ZEDLET_DIR}/state \ $ZED_LOG $ZED_DEBUG_LOG $VDEVID_CONF_ETC $VDEVID_CONF \ $ZEDLET_DIR } # # Check if ZED is currently running; if so, returns PIDs # function zed_check { if ! is_linux; then return fi zedpids="$(pgrep -x zed)" zedpids2="$(pgrep -x lt-zed)" echo ${zedpids} ${zedpids2} } # # Check if ZED is currently running, if not start ZED. # function zed_start { if ! is_linux; then return fi # ZEDLET_DIR=/var/tmp/zed if [[ ! -d $ZEDLET_DIR ]]; then log_must mkdir $ZEDLET_DIR fi # Verify the ZED is not already running. zedpids=$(zed_check) if [ -n "$zedpids" ]; then # We never, ever, really want it to just keep going if zed # is already running - usually this implies our test cases # will break very strangely because whatever we wanted to # configure zed for won't be listening to our changes in the # tmpdir log_fail "ZED already running - ${zedpids}" else log_note "Starting ZED" # run ZED in the background and redirect foreground logging # output to $ZED_LOG. log_must truncate -s 0 $ZED_DEBUG_LOG log_must eval "zed -vF -d $ZEDLET_DIR -P $PATH" \ "-s $ZEDLET_DIR/state -j 1 2>$ZED_LOG &" fi return 0 } # # Kill ZED process # function zed_stop { if ! is_linux; then return "" fi log_note "Stopping ZED" while true; do zedpids=$(zed_check) [ ! -n "$zedpids" ] && break log_must kill $zedpids sleep 1 done return 0 } # # Drain all zevents # function zed_events_drain { while [ $(zpool events -H | wc -l) -ne 0 ]; do sleep 1 zpool events -c >/dev/null done } # Set a variable in zed.rc to something, un-commenting it in the process. # # $1 variable # $2 value function zed_rc_set { var="$1" val="$2" # Remove the line cmd="'/$var/d'" eval sed -i $cmd $ZEDLET_DIR/zed.rc # Add it at the end echo "$var=$val" >> $ZEDLET_DIR/zed.rc } # # Check is provided device is being active used as a swap device. # function is_swap_inuse { typeset device=$1 if [[ -z $device ]] ; then log_note "No device specified." return 1 fi case "$UNAME" in Linux) swapon -s | grep -wq $(readlink -f $device) ;; FreeBSD) swapctl -l | grep -wq $device ;; *) swap -l | grep -wq $device ;; esac } # # Setup a swap device using the provided device. # function swap_setup { typeset swapdev=$1 case "$UNAME" in Linux) log_must eval "mkswap $swapdev > /dev/null 2>&1" log_must swapon $swapdev ;; FreeBSD) log_must swapctl -a $swapdev ;; *) log_must swap -a $swapdev ;; esac return 0 } # # Cleanup a swap device on the provided device. # function swap_cleanup { typeset swapdev=$1 if is_swap_inuse $swapdev; then if is_linux; then log_must swapoff $swapdev elif is_freebsd; then log_must swapoff $swapdev else log_must swap -d $swapdev fi fi return 0 } # # Set a global system tunable (64-bit value) # # $1 tunable name (use a NAME defined in tunables.cfg) # $2 tunable values # function set_tunable64 { set_tunable_impl "$1" "$2" Z } # # Set a global system tunable (32-bit value) # # $1 tunable name (use a NAME defined in tunables.cfg) # $2 tunable values # function set_tunable32 { set_tunable_impl "$1" "$2" W } function set_tunable_impl { typeset name="$1" typeset value="$2" typeset mdb_cmd="$3" eval "typeset tunable=\$$name" case "$tunable" in UNSUPPORTED) log_unsupported "Tunable '$name' is unsupported on $UNAME" ;; "") log_fail "Tunable '$name' must be added to tunables.cfg" ;; *) ;; esac [[ -z "$value" ]] && return 1 [[ -z "$mdb_cmd" ]] && return 1 case "$UNAME" in Linux) typeset zfs_tunables="/sys/module/zfs/parameters" echo "$value" >"$zfs_tunables/$tunable" ;; FreeBSD) sysctl vfs.zfs.$tunable=$value ;; SunOS) echo "${tunable}/${mdb_cmd}0t${value}" | mdb -kw ;; esac } # # Get a global system tunable # # $1 tunable name (use a NAME defined in tunables.cfg) # function get_tunable { get_tunable_impl "$1" } function get_tunable_impl { typeset name="$1" typeset module="${2:-zfs}" typeset check_only="$3" eval "typeset tunable=\$$name" case "$tunable" in UNSUPPORTED) if [ -z "$check_only" ] ; then log_unsupported "Tunable '$name' is unsupported on $UNAME" else return 1 fi ;; "") if [ -z "$check_only" ] ; then log_fail "Tunable '$name' must be added to tunables.cfg" else return 1 fi ;; *) ;; esac case "$UNAME" in Linux) typeset zfs_tunables="/sys/module/$module/parameters" cat $zfs_tunables/$tunable ;; FreeBSD) sysctl -n vfs.zfs.$tunable ;; SunOS) [[ "$module" -eq "zfs" ]] || return 1 ;; esac } # Does a tunable exist? # # $1: Tunable name function tunable_exists { get_tunable_impl $1 "zfs" 1 } # # Compute MD5 digest for given file or stdin if no file given. # Note: file path must not contain spaces # function md5digest { typeset file=$1 case "$UNAME" in FreeBSD) md5 -q $file ;; *) typeset sum _ read -r sum _ < <(md5sum -b $file) echo $sum ;; esac } # # Compute SHA256 digest for given file or stdin if no file given. # Note: file path must not contain spaces # function sha256digest { typeset file=$1 case "$UNAME" in FreeBSD) sha256 -q $file ;; *) typeset sum _ read -r sum _ < <(sha256sum -b $file) echo $sum ;; esac } function new_fs # { case "$UNAME" in FreeBSD) newfs "$@" ;; *) echo y | newfs -v "$@" ;; esac } function stat_size # { typeset path=$1 case "$UNAME" in FreeBSD) stat -f %z "$path" ;; *) stat -c %s "$path" ;; esac } function stat_mtime # { typeset path=$1 case "$UNAME" in FreeBSD) stat -f %m "$path" ;; *) stat -c %Y "$path" ;; esac } function stat_ctime # { typeset path=$1 case "$UNAME" in FreeBSD) stat -f %c "$path" ;; *) stat -c %Z "$path" ;; esac } function stat_crtime # { typeset path=$1 case "$UNAME" in FreeBSD) stat -f %B "$path" ;; *) stat -c %W "$path" ;; esac } function stat_generation # { typeset path=$1 case "$UNAME" in Linux) getversion "${path}" ;; *) stat -f %v "${path}" ;; esac } # Run a command as if it was being run in a TTY. # # Usage: # # faketty command # function faketty { if is_freebsd; then script -q /dev/null env "$@" else script --return --quiet -c "$*" /dev/null fi } # # Produce a random permutation of the integers in a given range (inclusive). # function range_shuffle # begin end { typeset -i begin=$1 typeset -i end=$2 seq ${begin} ${end} | sort -R } # # Cross-platform xattr helpers # function get_xattr # name path { typeset name=$1 typeset path=$2 case "$UNAME" in FreeBSD) getextattr -qq user "${name}" "${path}" ;; *) attr -qg "${name}" "${path}" ;; esac } function set_xattr # name value path { typeset name=$1 typeset value=$2 typeset path=$3 case "$UNAME" in FreeBSD) setextattr user "${name}" "${value}" "${path}" ;; *) attr -qs "${name}" -V "${value}" "${path}" ;; esac } function set_xattr_stdin # name value { typeset name=$1 typeset path=$2 case "$UNAME" in FreeBSD) setextattr -i user "${name}" "${path}" ;; *) attr -qs "${name}" "${path}" ;; esac } function rm_xattr # name path { typeset name=$1 typeset path=$2 case "$UNAME" in FreeBSD) rmextattr -q user "${name}" "${path}" ;; *) attr -qr "${name}" "${path}" ;; esac } function ls_xattr # path { typeset path=$1 case "$UNAME" in FreeBSD) lsextattr -qq user "${path}" ;; *) attr -ql "${path}" ;; esac } function kstat # stat flags? { typeset stat=$1 typeset flags=${2-"-n"} case "$UNAME" in FreeBSD) sysctl $flags kstat.zfs.misc.$stat ;; Linux) cat "/proc/spl/kstat/zfs/$stat" 2>/dev/null ;; *) false ;; esac } function get_arcstat # stat { typeset stat=$1 case "$UNAME" in FreeBSD) kstat arcstats.$stat ;; Linux) kstat arcstats | awk "/$stat/"' { print $3 }' ;; *) false ;; esac } function punch_hole # offset length file { typeset offset=$1 typeset length=$2 typeset file=$3 case "$UNAME" in FreeBSD) truncate -d -o $offset -l $length "$file" ;; Linux) fallocate --punch-hole --offset $offset --length $length "$file" ;; *) false ;; esac } function zero_range # offset length file { typeset offset=$1 typeset length=$2 typeset file=$3 case "$UNAME" in Linux) fallocate --zero-range --offset $offset --length $length "$file" ;; *) false ;; esac } # # Wait for the specified arcstat to reach non-zero quiescence. # If echo is 1 echo the value after reaching quiescence, otherwise # if echo is 0 print the arcstat we are waiting on. # function arcstat_quiescence # stat echo { typeset stat=$1 typeset echo=$2 typeset do_once=true if [[ $echo -eq 0 ]]; then echo "Waiting for arcstat $1 quiescence." fi while $do_once || [ $stat1 -ne $stat2 ] || [ $stat2 -eq 0 ]; do typeset stat1=$(get_arcstat $stat) sleep 2 typeset stat2=$(get_arcstat $stat) do_once=false done if [[ $echo -eq 1 ]]; then echo $stat2 fi } function arcstat_quiescence_noecho # stat { typeset stat=$1 arcstat_quiescence $stat 0 } function arcstat_quiescence_echo # stat { typeset stat=$1 arcstat_quiescence $stat 1 } # # Given an array of pids, wait until all processes # have completed and check their return status. # function wait_for_children #children { rv=0 children=("$@") for child in "${children[@]}" do child_exit=0 wait ${child} || child_exit=$? if [ $child_exit -ne 0 ]; then echo "child ${child} failed with ${child_exit}" rv=1 fi done return $rv } # # Compare two directory trees recursively in a manner similar to diff(1), but # using rsync. If there are any discrepancies, a summary of the differences are # output and a non-zero error is returned. # # If you're comparing a directory after a ZIL replay, you should set # LIBTEST_DIFF_ZIL_REPLAY=1 or use replay_directory_diff which will cause # directory_diff to ignore mtime changes (the ZIL replay won't fix up mtime # information). # function directory_diff # dir_a dir_b { dir_a="$1" dir_b="$2" zil_replay="${LIBTEST_DIFF_ZIL_REPLAY:-0}" # If one of the directories doesn't exist, return 2. This is to match the # semantics of diff. if ! [ -d "$dir_a" -a -d "$dir_b" ]; then return 2 fi # Run rsync with --dry-run --itemize-changes to get something akin to diff # output, but rsync is far more thorough in detecting differences (diff # doesn't compare file metadata, and cannot handle special files). # # Also make sure to filter out non-user.* xattrs when comparing. On # SELinux-enabled systems the copied tree will probably have different # SELinux labels. args=("-nicaAHX" '--filter=-x! user.*' "--delete") # NOTE: Quite a few rsync builds do not support --crtimes which would be # necessary to verify that creation times are being maintained properly. # Unfortunately because of this we cannot use it unconditionally but we can # check if this rsync build supports it and use it then. This check is # based on the same check in the rsync test suite (testsuite/crtimes.test). # # We check ctimes even with zil_replay=1 because the ZIL does store # creation times and we should make sure they match (if the creation times # do not match there is a "c" entry in one of the columns). if rsync --version | grep -q "[, ] crtimes"; then args+=("--crtimes") else log_note "This rsync package does not support --crtimes (-N)." fi # If we are testing a ZIL replay, we need to ignore timestamp changes. # Unfortunately --no-times doesn't do what we want -- it will still tell # you if the timestamps don't match but rsync will set the timestamps to # the current time (leading to an itemised change entry). It's simpler to # just filter out those lines. if [ "$zil_replay" -eq 0 ]; then filter=("cat") else # Different rsync versions have different numbers of columns. So just # require that aside from the first two, all other columns must be # blank (literal ".") or a timestamp field ("[tT]"). filter=("grep" "-v" '^\..[.Tt]\+ ') fi diff="$(rsync "${args[@]}" "$dir_a/" "$dir_b/" | "${filter[@]}")" rv=0 if [ -n "$diff" ]; then echo "$diff" rv=1 fi return $rv } # # Compare two directory trees recursively, without checking whether the mtimes # match (creation times will be checked if the available rsync binary supports # it). This is necessary for ZIL replay checks (because the ZIL does not # contain mtimes and thus after a ZIL replay, mtimes won't match). # # This is shorthand for LIBTEST_DIFF_ZIL_REPLAY=1 directory_diff <...>. # function replay_directory_diff # dir_a dir_b { LIBTEST_DIFF_ZIL_REPLAY=1 directory_diff "$@" } # # Put coredumps into $1/core.{basename} # # Output must be saved and passed to pop_coredump_pattern on cleanup # function push_coredump_pattern # dir { ulimit -c unlimited case "$UNAME" in Linux) cat /proc/sys/kernel/core_pattern /proc/sys/kernel/core_uses_pid echo "$1/core.%e" >/proc/sys/kernel/core_pattern && echo 0 >/proc/sys/kernel/core_uses_pid ;; FreeBSD) sysctl -n kern.corefile sysctl kern.corefile="$1/core.%N" >/dev/null ;; *) # Nothing to output – set only for this shell coreadm -p "$1/core.%f" ;; esac } # # Put coredumps back into the default location # function pop_coredump_pattern { [ -s "$1" ] || return 0 case "$UNAME" in Linux) typeset pat pid { read -r pat; read -r pid; } < "$1" echo "$pat" >/proc/sys/kernel/core_pattern && echo "$pid" >/proc/sys/kernel/core_uses_pid ;; FreeBSD) sysctl kern.corefile="$(<"$1")" >/dev/null ;; esac } diff --git a/tests/zfs-tests/tests/functional/fault/auto_offline_001_pos.ksh b/tests/zfs-tests/tests/functional/fault/auto_offline_001_pos.ksh index 17bde9a70636..0ab9317c0a06 100755 --- a/tests/zfs-tests/tests/functional/fault/auto_offline_001_pos.ksh +++ b/tests/zfs-tests/tests/functional/fault/auto_offline_001_pos.ksh @@ -1,188 +1,187 @@ #!/bin/ksh -p # # CDDL HEADER START # # This file and its contents are supplied under the terms of the # Common Development and Distribution License ("CDDL"), version 1.0. # You may only use this file in accordance with the terms of version # 1.0 of the CDDL. # # A full copy of the text of the CDDL should have accompanied this # source. A copy of the CDDL is also available via the Internet at # http://www.illumos.org/license/CDDL. # # CDDL HEADER END # # # Copyright 2018, loli10K . All rights reserved. # . $STF_SUITE/include/libtest.shlib . $STF_SUITE/tests/functional/events/events_common.kshlib . $STF_SUITE/tests/functional/fault/fault.cfg # # DESCRIPTION: -# Testing Fault Management Agent ZED Logic - Physically removed device is -# made unavail and onlined when reattached +# Testing Fault Management Agent ZED Logic - Physically detached device is +# made removed and onlined when reattached # # STRATEGY: # 1. Create a pool # 2. Simulate physical removal of one device -# 3. Verify the device is unavailable +# 3. Verify the device is removed when detached # 4. Reattach the device # 5. Verify the device is onlined # 6. Repeat the same tests with a spare device: # zed will use the spare to handle the removed data device # 7. Repeat the same tests again with a faulted spare device: -# the removed data device should be unavailable +# the removed data device should be removed # # NOTE: the use of 'block_device_wait' throughout the test helps avoid race # conditions caused by mixing creation/removal events from partitioning the # disk (zpool create) and events from physically removing it (remove_disk). # -# NOTE: the test relies on 'zpool sync' to prompt the kmods to transition a -# vdev to the unavailable state. The ZED does receive a removal notification -# but only relies on it to activate a hot spare. Additional work is planned -# to extend an existing ioctl interface to allow the ZED to transition the -# vdev in to a removed state. +# NOTE: the test relies on ZED to transit state to removed on device removed +# event. The ZED does receive a removal notification but only relies on it to +# activate a hot spare. Additional work is planned to extend an existing ioctl +# interface to allow the ZED to transition the vdev in to a removed state. # verify_runnable "both" if is_linux; then # Add one 512b scsi_debug device (4Kn would generate IO errors) # NOTE: must be larger than other "file" vdevs and minimum SPA devsize: # add 32m of fudge load_scsi_debug $(($MINVDEVSIZE/1024/1024+32)) 1 1 1 '512b' else log_unsupported "scsi debug module unsupported" fi function cleanup { destroy_pool $TESTPOOL rm -f $filedev1 rm -f $filedev2 rm -f $filedev3 rm -f $sparedev unload_scsi_debug } log_assert "ZED detects physically removed devices" log_onexit cleanup filedev1="$TEST_BASE_DIR/file-vdev-1" filedev2="$TEST_BASE_DIR/file-vdev-2" filedev3="$TEST_BASE_DIR/file-vdev-3" sparedev="$TEST_BASE_DIR/file-vdev-spare" removedev=$(get_debug_device) typeset poolconfs=( "mirror $filedev1 $removedev" "raidz3 $filedev1 $filedev2 $filedev3 $removedev" "mirror $filedev1 $filedev2 special mirror $filedev3 $removedev" ) log_must truncate -s $MINVDEVSIZE $filedev1 log_must truncate -s $MINVDEVSIZE $filedev2 log_must truncate -s $MINVDEVSIZE $filedev3 log_must truncate -s $MINVDEVSIZE $sparedev for conf in "${poolconfs[@]}" do # 1. Create a pool log_must zpool create -f $TESTPOOL $conf block_device_wait ${DEV_DSKDIR}/${removedev} mntpnt=$(get_prop mountpoint /$TESTPOOL) # 2. Simulate physical removal of one device remove_disk $removedev log_must mkfile 1m $mntpnt/file sync_pool $TESTPOOL - # 3. Verify the device is unavailable. - log_must wait_vdev_state $TESTPOOL $removedev "UNAVAIL" + # 3. Verify the device is removed. + log_must wait_vdev_state $TESTPOOL $removedev "REMOVED" # 4. Reattach the device insert_disk $removedev # 5. Verify the device is onlined log_must wait_vdev_state $TESTPOOL $removedev "ONLINE" # cleanup destroy_pool $TESTPOOL log_must parted "${DEV_DSKDIR}/${removedev}" -s -- mklabel msdos block_device_wait ${DEV_DSKDIR}/${removedev} done # 6. Repeat the same tests with a spare device: zed will use the spare to handle # the removed data device for conf in "${poolconfs[@]}" do # 1. Create a pool with a spare log_must zpool create -f $TESTPOOL $conf block_device_wait ${DEV_DSKDIR}/${removedev} log_must zpool add $TESTPOOL spare $sparedev mntpnt=$(get_prop mountpoint /$TESTPOOL) # 2. Simulate physical removal of one device remove_disk $removedev log_must mkfile 1m $mntpnt/file sync_pool $TESTPOOL # 3. Verify the device is handled by the spare. log_must wait_hotspare_state $TESTPOOL $sparedev "INUSE" - log_must wait_vdev_state $TESTPOOL $removedev "UNAVAIL" + log_must wait_vdev_state $TESTPOOL $removedev "REMOVED" # 4. Reattach the device insert_disk $removedev # 5. Verify the device is onlined log_must wait_vdev_state $TESTPOOL $removedev "ONLINE" # cleanup destroy_pool $TESTPOOL log_must parted "${DEV_DSKDIR}/${removedev}" -s -- mklabel msdos block_device_wait ${DEV_DSKDIR}/${removedev} done # 7. Repeat the same tests again with a faulted spare device: zed should offline # the removed data device if no spare is available for conf in "${poolconfs[@]}" do # 1. Create a pool with a spare log_must zpool create -f $TESTPOOL $conf block_device_wait ${DEV_DSKDIR}/${removedev} log_must zpool add $TESTPOOL spare $sparedev mntpnt=$(get_prop mountpoint /$TESTPOOL) # 2. Fault the spare device making it unavailable log_must zpool offline -f $TESTPOOL $sparedev log_must wait_hotspare_state $TESTPOOL $sparedev "FAULTED" # 3. Simulate physical removal of one device remove_disk $removedev log_must mkfile 1m $mntpnt/file sync_pool $TESTPOOL - # 4. Verify the device is unavailable - log_must wait_vdev_state $TESTPOOL $removedev "UNAVAIL" + # 4. Verify the device is removed + log_must wait_vdev_state $TESTPOOL $removedev "REMOVED" # 5. Reattach the device insert_disk $removedev # 6. Verify the device is onlined log_must wait_vdev_state $TESTPOOL $removedev "ONLINE" # cleanup destroy_pool $TESTPOOL log_must parted "${DEV_DSKDIR}/${removedev}" -s -- mklabel msdos block_device_wait ${DEV_DSKDIR}/${removedev} done log_pass "ZED detects physically removed devices"