diff options
author | Brian Behlendorf <[email protected]> | 2008-12-11 11:08:09 -0800 |
---|---|---|
committer | Brian Behlendorf <[email protected]> | 2008-12-11 11:08:09 -0800 |
commit | 172bb4bd5e4afef721dd4d2972d8680d983f144b (patch) | |
tree | 18ab1e97e5e409150066c529b5a981ecf600ef80 /cmd/zpool | |
parent | 9e8b1e836caa454586797f771a7ad1817ebae315 (diff) |
Move the world out of /zfs/ and seperate out module build tree
Diffstat (limited to 'cmd/zpool')
-rw-r--r-- | cmd/zpool/zpool_iter.c | 252 | ||||
-rw-r--r-- | cmd/zpool/zpool_main.c | 3950 | ||||
-rw-r--r-- | cmd/zpool/zpool_util.c | 104 | ||||
-rw-r--r-- | cmd/zpool/zpool_util.h | 72 | ||||
-rw-r--r-- | cmd/zpool/zpool_vdev.c | 1401 |
5 files changed, 5779 insertions, 0 deletions
diff --git a/cmd/zpool/zpool_iter.c b/cmd/zpool/zpool_iter.c new file mode 100644 index 000000000..2f0daefd5 --- /dev/null +++ b/cmd/zpool/zpool_iter.c @@ -0,0 +1,252 @@ +/* + * 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 http://www.opensolaris.org/os/licensing. + * 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 2007 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <libintl.h> +#include <libuutil.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> + +#include <libzfs.h> + +#include "zpool_util.h" + +/* + * Private interface for iterating over pools specified on the command line. + * Most consumers will call for_each_pool, but in order to support iostat, we + * allow fined grained control through the zpool_list_t interface. + */ + +typedef struct zpool_node { + zpool_handle_t *zn_handle; + uu_avl_node_t zn_avlnode; + int zn_mark; +} zpool_node_t; + +struct zpool_list { + boolean_t zl_findall; + uu_avl_t *zl_avl; + uu_avl_pool_t *zl_pool; + zprop_list_t **zl_proplist; +}; + +/* ARGSUSED */ +static int +zpool_compare(const void *larg, const void *rarg, void *unused) +{ + zpool_handle_t *l = ((zpool_node_t *)larg)->zn_handle; + zpool_handle_t *r = ((zpool_node_t *)rarg)->zn_handle; + const char *lname = zpool_get_name(l); + const char *rname = zpool_get_name(r); + + return (strcmp(lname, rname)); +} + +/* + * Callback function for pool_list_get(). Adds the given pool to the AVL tree + * of known pools. + */ +static int +add_pool(zpool_handle_t *zhp, void *data) +{ + zpool_list_t *zlp = data; + zpool_node_t *node = safe_malloc(sizeof (zpool_node_t)); + uu_avl_index_t idx; + + node->zn_handle = zhp; + uu_avl_node_init(node, &node->zn_avlnode, zlp->zl_pool); + if (uu_avl_find(zlp->zl_avl, node, NULL, &idx) == NULL) { + if (zlp->zl_proplist && + zpool_expand_proplist(zhp, zlp->zl_proplist) != 0) { + zpool_close(zhp); + free(node); + return (-1); + } + uu_avl_insert(zlp->zl_avl, node, idx); + } else { + zpool_close(zhp); + free(node); + return (-1); + } + + return (0); +} + +/* + * Create a list of pools based on the given arguments. If we're given no + * arguments, then iterate over all pools in the system and add them to the AVL + * tree. Otherwise, add only those pool explicitly specified on the command + * line. + */ +zpool_list_t * +pool_list_get(int argc, char **argv, zprop_list_t **proplist, int *err) +{ + zpool_list_t *zlp; + + zlp = safe_malloc(sizeof (zpool_list_t)); + + zlp->zl_pool = uu_avl_pool_create("zfs_pool", sizeof (zpool_node_t), + offsetof(zpool_node_t, zn_avlnode), zpool_compare, UU_DEFAULT); + + if (zlp->zl_pool == NULL) + zpool_no_memory(); + + if ((zlp->zl_avl = uu_avl_create(zlp->zl_pool, NULL, + UU_DEFAULT)) == NULL) + zpool_no_memory(); + + zlp->zl_proplist = proplist; + + if (argc == 0) { + (void) zpool_iter(g_zfs, add_pool, zlp); + zlp->zl_findall = B_TRUE; + } else { + int i; + + for (i = 0; i < argc; i++) { + zpool_handle_t *zhp; + + if (zhp = zpool_open_canfail(g_zfs, argv[i])) { + if (add_pool(zhp, zlp) != 0) + *err = B_TRUE; + } else { + *err = B_TRUE; + } + } + } + + return (zlp); +} + +/* + * Search for any new pools, adding them to the list. We only add pools when no + * options were given on the command line. Otherwise, we keep the list fixed as + * those that were explicitly specified. + */ +void +pool_list_update(zpool_list_t *zlp) +{ + if (zlp->zl_findall) + (void) zpool_iter(g_zfs, add_pool, zlp); +} + +/* + * Iterate over all pools in the list, executing the callback for each + */ +int +pool_list_iter(zpool_list_t *zlp, int unavail, zpool_iter_f func, + void *data) +{ + zpool_node_t *node, *next_node; + int ret = 0; + + for (node = uu_avl_first(zlp->zl_avl); node != NULL; node = next_node) { + next_node = uu_avl_next(zlp->zl_avl, node); + if (zpool_get_state(node->zn_handle) != POOL_STATE_UNAVAIL || + unavail) + ret |= func(node->zn_handle, data); + } + + return (ret); +} + +/* + * Remove the given pool from the list. When running iostat, we want to remove + * those pools that no longer exist. + */ +void +pool_list_remove(zpool_list_t *zlp, zpool_handle_t *zhp) +{ + zpool_node_t search, *node; + + search.zn_handle = zhp; + if ((node = uu_avl_find(zlp->zl_avl, &search, NULL, NULL)) != NULL) { + uu_avl_remove(zlp->zl_avl, node); + zpool_close(node->zn_handle); + free(node); + } +} + +/* + * Free all the handles associated with this list. + */ +void +pool_list_free(zpool_list_t *zlp) +{ + uu_avl_walk_t *walk; + zpool_node_t *node; + + if ((walk = uu_avl_walk_start(zlp->zl_avl, UU_WALK_ROBUST)) == NULL) { + (void) fprintf(stderr, + gettext("internal error: out of memory")); + exit(1); + } + + while ((node = uu_avl_walk_next(walk)) != NULL) { + uu_avl_remove(zlp->zl_avl, node); + zpool_close(node->zn_handle); + free(node); + } + + uu_avl_walk_end(walk); + uu_avl_destroy(zlp->zl_avl); + uu_avl_pool_destroy(zlp->zl_pool); + + free(zlp); +} + +/* + * Returns the number of elements in the pool list. + */ +int +pool_list_count(zpool_list_t *zlp) +{ + return (uu_avl_numnodes(zlp->zl_avl)); +} + +/* + * High level function which iterates over all pools given on the command line, + * using the pool_list_* interfaces. + */ +int +for_each_pool(int argc, char **argv, boolean_t unavail, + zprop_list_t **proplist, zpool_iter_f func, void *data) +{ + zpool_list_t *list; + int ret = 0; + + if ((list = pool_list_get(argc, argv, proplist, &ret)) == NULL) + return (1); + + if (pool_list_iter(list, unavail, func, data) != 0) + ret = 1; + + pool_list_free(list); + + return (ret); +} diff --git a/cmd/zpool/zpool_main.c b/cmd/zpool/zpool_main.c new file mode 100644 index 000000000..54bba8645 --- /dev/null +++ b/cmd/zpool/zpool_main.c @@ -0,0 +1,3950 @@ +/* + * 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 http://www.opensolaris.org/os/licensing. + * 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 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#include <assert.h> +#include <ctype.h> +#include <dirent.h> +#include <errno.h> +#include <fcntl.h> +#include <libgen.h> +#include <libintl.h> +#include <libuutil.h> +#include <locale.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <strings.h> +#include <unistd.h> +#include <priv.h> +#include <pwd.h> +#include <zone.h> +#include <sys/fs/zfs.h> + +#include <sys/stat.h> + +#include <libzfs.h> + +#include "zpool_util.h" +#include "zfs_comutil.h" + +static int zpool_do_create(int, char **); +static int zpool_do_destroy(int, char **); + +static int zpool_do_add(int, char **); +static int zpool_do_remove(int, char **); + +static int zpool_do_list(int, char **); +static int zpool_do_iostat(int, char **); +static int zpool_do_status(int, char **); + +static int zpool_do_online(int, char **); +static int zpool_do_offline(int, char **); +static int zpool_do_clear(int, char **); + +static int zpool_do_attach(int, char **); +static int zpool_do_detach(int, char **); +static int zpool_do_replace(int, char **); + +static int zpool_do_scrub(int, char **); + +static int zpool_do_import(int, char **); +static int zpool_do_export(int, char **); + +static int zpool_do_upgrade(int, char **); + +static int zpool_do_history(int, char **); + +static int zpool_do_get(int, char **); +static int zpool_do_set(int, char **); + +/* + * These libumem hooks provide a reasonable set of defaults for the allocator's + * debugging facilities. + */ + +#ifdef DEBUG +const char * +_umem_debug_init(void) +{ + return ("default,verbose"); /* $UMEM_DEBUG setting */ +} + +const char * +_umem_logging_init(void) +{ + return ("fail,contents"); /* $UMEM_LOGGING setting */ +} +#endif + +typedef enum { + HELP_ADD, + HELP_ATTACH, + HELP_CLEAR, + HELP_CREATE, + HELP_DESTROY, + HELP_DETACH, + HELP_EXPORT, + HELP_HISTORY, + HELP_IMPORT, + HELP_IOSTAT, + HELP_LIST, + HELP_OFFLINE, + HELP_ONLINE, + HELP_REPLACE, + HELP_REMOVE, + HELP_SCRUB, + HELP_STATUS, + HELP_UPGRADE, + HELP_GET, + HELP_SET +} zpool_help_t; + + +typedef struct zpool_command { + const char *name; + int (*func)(int, char **); + zpool_help_t usage; +} zpool_command_t; + +/* + * Master command table. Each ZFS command has a name, associated function, and + * usage message. The usage messages need to be internationalized, so we have + * to have a function to return the usage message based on a command index. + * + * These commands are organized according to how they are displayed in the usage + * message. An empty command (one with a NULL name) indicates an empty line in + * the generic usage message. + */ +static zpool_command_t command_table[] = { + { "create", zpool_do_create, HELP_CREATE }, + { "destroy", zpool_do_destroy, HELP_DESTROY }, + { NULL }, + { "add", zpool_do_add, HELP_ADD }, + { "remove", zpool_do_remove, HELP_REMOVE }, + { NULL }, + { "list", zpool_do_list, HELP_LIST }, + { "iostat", zpool_do_iostat, HELP_IOSTAT }, + { "status", zpool_do_status, HELP_STATUS }, + { NULL }, + { "online", zpool_do_online, HELP_ONLINE }, + { "offline", zpool_do_offline, HELP_OFFLINE }, + { "clear", zpool_do_clear, HELP_CLEAR }, + { NULL }, + { "attach", zpool_do_attach, HELP_ATTACH }, + { "detach", zpool_do_detach, HELP_DETACH }, + { "replace", zpool_do_replace, HELP_REPLACE }, + { NULL }, + { "scrub", zpool_do_scrub, HELP_SCRUB }, + { NULL }, + { "import", zpool_do_import, HELP_IMPORT }, + { "export", zpool_do_export, HELP_EXPORT }, + { "upgrade", zpool_do_upgrade, HELP_UPGRADE }, + { NULL }, + { "history", zpool_do_history, HELP_HISTORY }, + { "get", zpool_do_get, HELP_GET }, + { "set", zpool_do_set, HELP_SET }, +}; + +#define NCOMMAND (sizeof (command_table) / sizeof (command_table[0])) + +zpool_command_t *current_command; +static char history_str[HIS_MAX_RECORD_LEN]; + +static const char * +get_usage(zpool_help_t idx) { + switch (idx) { + case HELP_ADD: + return (gettext("\tadd [-fn] <pool> <vdev> ...\n")); + case HELP_ATTACH: + return (gettext("\tattach [-f] <pool> <device> " + "<new-device>\n")); + case HELP_CLEAR: + return (gettext("\tclear <pool> [device]\n")); + case HELP_CREATE: + return (gettext("\tcreate [-fn] [-o property=value] ... \n" + "\t [-O file-system-property=value] ... \n" + "\t [-m mountpoint] [-R root] <pool> <vdev> ...\n")); + case HELP_DESTROY: + return (gettext("\tdestroy [-f] <pool>\n")); + case HELP_DETACH: + return (gettext("\tdetach <pool> <device>\n")); + case HELP_EXPORT: + return (gettext("\texport [-f] <pool> ...\n")); + case HELP_HISTORY: + return (gettext("\thistory [-il] [<pool>] ...\n")); + case HELP_IMPORT: + return (gettext("\timport [-d dir] [-D]\n" + "\timport [-o mntopts] [-o property=value] ... \n" + "\t [-d dir | -c cachefile] [-D] [-f] [-R root] -a\n" + "\timport [-o mntopts] [-o property=value] ... \n" + "\t [-d dir | -c cachefile] [-D] [-f] [-R root] " + "<pool | id> [newpool]\n")); + case HELP_IOSTAT: + return (gettext("\tiostat [-v] [pool] ... [interval " + "[count]]\n")); + case HELP_LIST: + return (gettext("\tlist [-H] [-o property[,...]] " + "[pool] ...\n")); + case HELP_OFFLINE: + return (gettext("\toffline [-t] <pool> <device> ...\n")); + case HELP_ONLINE: + return (gettext("\tonline <pool> <device> ...\n")); + case HELP_REPLACE: + return (gettext("\treplace [-f] <pool> <device> " + "[new-device]\n")); + case HELP_REMOVE: + return (gettext("\tremove <pool> <device> ...\n")); + case HELP_SCRUB: + return (gettext("\tscrub [-s] <pool> ...\n")); + case HELP_STATUS: + return (gettext("\tstatus [-vx] [pool] ...\n")); + case HELP_UPGRADE: + return (gettext("\tupgrade\n" + "\tupgrade -v\n" + "\tupgrade [-V version] <-a | pool ...>\n")); + case HELP_GET: + return (gettext("\tget <\"all\" | property[,...]> " + "<pool> ...\n")); + case HELP_SET: + return (gettext("\tset <property=value> <pool> \n")); + } + + abort(); + /* NOTREACHED */ +} + + +/* + * Callback routine that will print out a pool property value. + */ +static int +print_prop_cb(int prop, void *cb) +{ + FILE *fp = cb; + + (void) fprintf(fp, "\t%-13s ", zpool_prop_to_name(prop)); + + if (zpool_prop_readonly(prop)) + (void) fprintf(fp, " NO "); + else + (void) fprintf(fp, " YES "); + + if (zpool_prop_values(prop) == NULL) + (void) fprintf(fp, "-\n"); + else + (void) fprintf(fp, "%s\n", zpool_prop_values(prop)); + + return (ZPROP_CONT); +} + +/* + * Display usage message. If we're inside a command, display only the usage for + * that command. Otherwise, iterate over the entire command table and display + * a complete usage message. + */ +void +usage(boolean_t requested) +{ + FILE *fp = requested ? stdout : stderr; + + if (current_command == NULL) { + int i; + + (void) fprintf(fp, gettext("usage: zpool command args ...\n")); + (void) fprintf(fp, + gettext("where 'command' is one of the following:\n\n")); + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + (void) fprintf(fp, "\n"); + else + (void) fprintf(fp, "%s", + get_usage(command_table[i].usage)); + } + } else { + (void) fprintf(fp, gettext("usage:\n")); + (void) fprintf(fp, "%s", get_usage(current_command->usage)); + } + + if (current_command != NULL && + ((strcmp(current_command->name, "set") == 0) || + (strcmp(current_command->name, "get") == 0) || + (strcmp(current_command->name, "list") == 0))) { + + (void) fprintf(fp, + gettext("\nthe following properties are supported:\n")); + + (void) fprintf(fp, "\n\t%-13s %s %s\n\n", + "PROPERTY", "EDIT", "VALUES"); + + /* Iterate over all properties */ + (void) zprop_iter(print_prop_cb, fp, B_FALSE, B_TRUE, + ZFS_TYPE_POOL); + } + + /* + * See comments at end of main(). + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + exit(requested ? 0 : 2); +} + +void +print_vdev_tree(zpool_handle_t *zhp, const char *name, nvlist_t *nv, int indent, + boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + char *vname; + + if (name != NULL) + (void) printf("\t%*s%s\n", indent, "", name); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return; + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + + vname = zpool_vdev_name(g_zfs, zhp, child[c]); + print_vdev_tree(zhp, vname, child[c], indent + 2, + B_FALSE); + free(vname); + } +} + +/* + * Add a property pair (name, string-value) into a property nvlist. + */ +static int +add_prop_list(const char *propname, char *propval, nvlist_t **props, + boolean_t poolprop) +{ + zpool_prop_t prop = ZPROP_INVAL; + zfs_prop_t fprop; + nvlist_t *proplist; + const char *normnm; + char *strval; + + if (*props == NULL && + nvlist_alloc(props, NV_UNIQUE_NAME, 0) != 0) { + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + return (1); + } + + proplist = *props; + + if (poolprop) { + if ((prop = zpool_name_to_prop(propname)) == ZPROP_INVAL) { + (void) fprintf(stderr, gettext("property '%s' is " + "not a valid pool property\n"), propname); + return (2); + } + normnm = zpool_prop_to_name(prop); + } else { + if ((fprop = zfs_name_to_prop(propname)) == ZPROP_INVAL) { + (void) fprintf(stderr, gettext("property '%s' is " + "not a valid file system property\n"), propname); + return (2); + } + normnm = zfs_prop_to_name(fprop); + } + + if (nvlist_lookup_string(proplist, normnm, &strval) == 0 && + prop != ZPOOL_PROP_CACHEFILE) { + (void) fprintf(stderr, gettext("property '%s' " + "specified multiple times\n"), propname); + return (2); + } + + if (nvlist_add_string(proplist, normnm, propval) != 0) { + (void) fprintf(stderr, gettext("internal " + "error: out of memory\n")); + return (1); + } + + return (0); +} + +/* + * zpool add [-fn] <pool> <vdev> ... + * + * -f Force addition of devices, even if they appear in use + * -n Do not add the devices, but display the resulting layout if + * they were to be added. + * + * Adds the given vdevs to 'pool'. As with create, the bulk of this work is + * handled by get_vdev_spec(), which constructs the nvlist needed to pass to + * libzfs. + */ +int +zpool_do_add(int argc, char **argv) +{ + boolean_t force = B_FALSE; + boolean_t dryrun = B_FALSE; + int c; + nvlist_t *nvroot; + char *poolname; + int ret; + zpool_handle_t *zhp; + nvlist_t *config; + + /* check options */ + while ((c = getopt(argc, argv, "fn")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case 'n': + dryrun = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing vdev specification\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + argc--; + argv++; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + if ((config = zpool_get_config(zhp, NULL)) == NULL) { + (void) fprintf(stderr, gettext("pool '%s' is unavailable\n"), + poolname); + zpool_close(zhp); + return (1); + } + + /* pass off to get_vdev_spec for processing */ + nvroot = make_root_vdev(zhp, force, !force, B_FALSE, dryrun, + argc, argv); + if (nvroot == NULL) { + zpool_close(zhp); + return (1); + } + + if (dryrun) { + nvlist_t *poolnvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &poolnvroot) == 0); + + (void) printf(gettext("would update '%s' to the following " + "configuration:\n"), zpool_get_name(zhp)); + + /* print original main pool and new tree */ + print_vdev_tree(zhp, poolname, poolnvroot, 0, B_FALSE); + print_vdev_tree(zhp, NULL, nvroot, 0, B_FALSE); + + /* Do the same for the logs */ + if (num_logs(poolnvroot) > 0) { + print_vdev_tree(zhp, "logs", poolnvroot, 0, B_TRUE); + print_vdev_tree(zhp, NULL, nvroot, 0, B_TRUE); + } else if (num_logs(nvroot) > 0) { + print_vdev_tree(zhp, "logs", nvroot, 0, B_TRUE); + } + + ret = 0; + } else { + ret = (zpool_add(zhp, nvroot) != 0); + } + + nvlist_free(nvroot); + zpool_close(zhp); + + return (ret); +} + +/* + * zpool remove <pool> <vdev> ... + * + * Removes the given vdev from the pool. Currently, this only supports removing + * spares and cache devices from the pool. Eventually, we'll want to support + * removing leaf vdevs (as an alias for 'detach') as well as toplevel vdevs. + */ +int +zpool_do_remove(int argc, char **argv) +{ + char *poolname; + int i, ret = 0; + zpool_handle_t *zhp; + + argc--; + argv++; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_remove(zhp, argv[i]) != 0) + ret = 1; + } + + return (ret); +} + +/* + * zpool create [-fn] [-o property=value] ... + * [-O file-system-property=value] ... + * [-R root] [-m mountpoint] <pool> <dev> ... + * + * -f Force creation, even if devices appear in use + * -n Do not create the pool, but display the resulting layout if it + * were to be created. + * -R Create a pool under an alternate root + * -m Set default mountpoint for the root dataset. By default it's + * '/<pool>' + * -o Set property=value. + * -O Set fsproperty=value in the pool's root file system + * + * Creates the named pool according to the given vdev specification. The + * bulk of the vdev processing is done in get_vdev_spec() in zpool_vdev.c. Once + * we get the nvlist back from get_vdev_spec(), we either print out the contents + * (if '-n' was specified), or pass it to libzfs to do the creation. + */ +int +zpool_do_create(int argc, char **argv) +{ + boolean_t force = B_FALSE; + boolean_t dryrun = B_FALSE; + int c; + nvlist_t *nvroot = NULL; + char *poolname; + int ret = 1; + char *altroot = NULL; + char *mountpoint = NULL; + nvlist_t *fsprops = NULL; + nvlist_t *props = NULL; + char *propval; + + /* check options */ + while ((c = getopt(argc, argv, ":fnR:m:o:O:")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case 'n': + dryrun = B_TRUE; + break; + case 'R': + altroot = optarg; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE)) + goto errout; + if (nvlist_lookup_string(props, + zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), + &propval) == 0) + break; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE)) + goto errout; + break; + case 'm': + mountpoint = optarg; + break; + case 'o': + if ((propval = strchr(optarg, '=')) == NULL) { + (void) fprintf(stderr, gettext("missing " + "'=' for -o option\n")); + goto errout; + } + *propval = '\0'; + propval++; + + if (add_prop_list(optarg, propval, &props, B_TRUE)) + goto errout; + break; + case 'O': + if ((propval = strchr(optarg, '=')) == NULL) { + (void) fprintf(stderr, gettext("missing " + "'=' for -O option\n")); + goto errout; + } + *propval = '\0'; + propval++; + + if (add_prop_list(optarg, propval, &fsprops, B_FALSE)) + goto errout; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + goto badusage; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + goto badusage; + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + goto badusage; + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing vdev specification\n")); + goto badusage; + } + + poolname = argv[0]; + + /* + * As a special case, check for use of '/' in the name, and direct the + * user to use 'zfs create' instead. + */ + if (strchr(poolname, '/') != NULL) { + (void) fprintf(stderr, gettext("cannot create '%s': invalid " + "character '/' in pool name\n"), poolname); + (void) fprintf(stderr, gettext("use 'zfs create' to " + "create a dataset\n")); + goto errout; + } + + /* pass off to get_vdev_spec for bulk processing */ + nvroot = make_root_vdev(NULL, force, !force, B_FALSE, dryrun, + argc - 1, argv + 1); + if (nvroot == NULL) + goto errout; + + /* make_root_vdev() allows 0 toplevel children if there are spares */ + if (!zfs_allocatable_devs(nvroot)) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: at least one toplevel vdev must be " + "specified\n")); + goto errout; + } + + + if (altroot != NULL && altroot[0] != '/') { + (void) fprintf(stderr, gettext("invalid alternate root '%s': " + "must be an absolute path\n"), altroot); + goto errout; + } + + /* + * Check the validity of the mountpoint and direct the user to use the + * '-m' mountpoint option if it looks like its in use. + */ + if (mountpoint == NULL || + (strcmp(mountpoint, ZFS_MOUNTPOINT_LEGACY) != 0 && + strcmp(mountpoint, ZFS_MOUNTPOINT_NONE) != 0)) { + char buf[MAXPATHLEN]; + DIR *dirp; + + if (mountpoint && mountpoint[0] != '/') { + (void) fprintf(stderr, gettext("invalid mountpoint " + "'%s': must be an absolute path, 'legacy', or " + "'none'\n"), mountpoint); + goto errout; + } + + if (mountpoint == NULL) { + if (altroot != NULL) + (void) snprintf(buf, sizeof (buf), "%s/%s", + altroot, poolname); + else + (void) snprintf(buf, sizeof (buf), "/%s", + poolname); + } else { + if (altroot != NULL) + (void) snprintf(buf, sizeof (buf), "%s%s", + altroot, mountpoint); + else + (void) snprintf(buf, sizeof (buf), "%s", + mountpoint); + } + + if ((dirp = opendir(buf)) == NULL && errno != ENOENT) { + (void) fprintf(stderr, gettext("mountpoint '%s' : " + "%s\n"), buf, strerror(errno)); + (void) fprintf(stderr, gettext("use '-m' " + "option to provide a different default\n")); + goto errout; + } else if (dirp) { + int count = 0; + + while (count < 3 && readdir(dirp) != NULL) + count++; + (void) closedir(dirp); + + if (count > 2) { + (void) fprintf(stderr, gettext("mountpoint " + "'%s' exists and is not empty\n"), buf); + (void) fprintf(stderr, gettext("use '-m' " + "option to provide a " + "different default\n")); + goto errout; + } + } + } + + if (dryrun) { + /* + * For a dry run invocation, print out a basic message and run + * through all the vdevs in the list and print out in an + * appropriate hierarchy. + */ + (void) printf(gettext("would create '%s' with the " + "following layout:\n\n"), poolname); + + print_vdev_tree(NULL, poolname, nvroot, 0, B_FALSE); + if (num_logs(nvroot) > 0) + print_vdev_tree(NULL, "logs", nvroot, 0, B_TRUE); + + ret = 0; + } else { + /* + * Hand off to libzfs. + */ + if (zpool_create(g_zfs, poolname, + nvroot, props, fsprops) == 0) { + zfs_handle_t *pool = zfs_open(g_zfs, poolname, + ZFS_TYPE_FILESYSTEM); + if (pool != NULL) { + if (mountpoint != NULL) + verify(zfs_prop_set(pool, + zfs_prop_to_name( + ZFS_PROP_MOUNTPOINT), + mountpoint) == 0); + if (zfs_mount(pool, NULL, 0) == 0) + ret = zfs_shareall(pool); + zfs_close(pool); + } + } else if (libzfs_errno(g_zfs) == EZFS_INVALIDNAME) { + (void) fprintf(stderr, gettext("pool name may have " + "been omitted\n")); + } + } + +errout: + nvlist_free(nvroot); + nvlist_free(fsprops); + nvlist_free(props); + return (ret); +badusage: + nvlist_free(fsprops); + nvlist_free(props); + usage(B_FALSE); + return (2); +} + +/* + * zpool destroy <pool> + * + * -f Forcefully unmount any datasets + * + * Destroy the given pool. Automatically unmounts any datasets in the pool. + */ +int +zpool_do_destroy(int argc, char **argv) +{ + boolean_t force = B_FALSE; + int c; + char *pool; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool argument\n")); + usage(B_FALSE); + } + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + pool = argv[0]; + + if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) { + /* + * As a special case, check for use of '/' in the name, and + * direct the user to use 'zfs destroy' instead. + */ + if (strchr(pool, '/') != NULL) + (void) fprintf(stderr, gettext("use 'zfs destroy' to " + "destroy a dataset\n")); + return (1); + } + + if (zpool_disable_datasets(zhp, force) != 0) { + (void) fprintf(stderr, gettext("could not destroy '%s': " + "could not unmount datasets\n"), zpool_get_name(zhp)); + return (1); + } + + ret = (zpool_destroy(zhp) != 0); + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool export [-f] <pool> ... + * + * -f Forcefully unmount datasets + * + * Export the given pools. By default, the command will attempt to cleanly + * unmount any active datasets within the pool. If the '-f' flag is specified, + * then the datasets will be forcefully unmounted. + */ +int +zpool_do_export(int argc, char **argv) +{ + boolean_t force = B_FALSE; + int c; + zpool_handle_t *zhp; + int ret; + int i; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* check arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool argument\n")); + usage(B_FALSE); + } + + ret = 0; + for (i = 0; i < argc; i++) { + if ((zhp = zpool_open_canfail(g_zfs, argv[i])) == NULL) { + ret = 1; + continue; + } + + if (zpool_disable_datasets(zhp, force) != 0) { + ret = 1; + zpool_close(zhp); + continue; + } + + if (zpool_export(zhp, force) != 0) + ret = 1; + + zpool_close(zhp); + } + + return (ret); +} + +/* + * Given a vdev configuration, determine the maximum width needed for the device + * name column. + */ +static int +max_width(zpool_handle_t *zhp, nvlist_t *nv, int depth, int max) +{ + char *name = zpool_vdev_name(g_zfs, zhp, nv); + nvlist_t **child; + uint_t c, children; + int ret; + + if (strlen(name) + depth > max) + max = strlen(name) + depth; + + free(name); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if ((ret = max_width(zhp, child[c], depth + 2, + max)) > max) + max = ret; + } + + + return (max); +} + + +/* + * Print the configuration of an exported pool. Iterate over all vdevs in the + * pool, printing out the name and status for each one. + */ +void +print_import_config(const char *name, nvlist_t *nv, int namewidth, int depth, + boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + vdev_stat_t *vs; + char *type, *vname; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + if (strcmp(type, VDEV_TYPE_MISSING) == 0) + return; + + verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + + (void) printf("\t%*s%-*s", depth, "", namewidth - depth, name); + (void) printf(" %s", zpool_state_to_name(vs->vs_state, vs->vs_aux)); + + if (vs->vs_aux != 0) { + (void) printf(" "); + + switch (vs->vs_aux) { + case VDEV_AUX_OPEN_FAILED: + (void) printf(gettext("cannot open")); + break; + + case VDEV_AUX_BAD_GUID_SUM: + (void) printf(gettext("missing device")); + break; + + case VDEV_AUX_NO_REPLICAS: + (void) printf(gettext("insufficient replicas")); + break; + + case VDEV_AUX_VERSION_NEWER: + (void) printf(gettext("newer version")); + break; + + case VDEV_AUX_ERR_EXCEEDED: + (void) printf(gettext("too many errors")); + break; + + default: + (void) printf(gettext("corrupted data")); + break; + } + } + (void) printf("\n"); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return; + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + print_import_config(vname, child[c], + namewidth, depth + 2, B_FALSE); + free(vname); + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) { + (void) printf(gettext("\tcache\n")); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + (void) printf("\t %s\n", vname); + free(vname); + } + } + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) { + (void) printf(gettext("\tspares\n")); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, NULL, child[c]); + (void) printf("\t %s\n", vname); + free(vname); + } + } +} + +/* + * Display the status for the given pool. + */ +static void +show_import(nvlist_t *config) +{ + uint64_t pool_state; + vdev_stat_t *vs; + char *name; + uint64_t guid; + char *msgid; + nvlist_t *nvroot; + int reason; + const char *health; + uint_t vsc; + int namewidth; + + verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, + &name) == 0); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, + &guid) == 0); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, + &pool_state) == 0); + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &vsc) == 0); + health = zpool_state_to_name(vs->vs_state, vs->vs_aux); + + reason = zpool_import_status(config, &msgid); + + (void) printf(gettext(" pool: %s\n"), name); + (void) printf(gettext(" id: %llu\n"), (u_longlong_t)guid); + (void) printf(gettext(" state: %s"), health); + if (pool_state == POOL_STATE_DESTROYED) + (void) printf(gettext(" (DESTROYED)")); + (void) printf("\n"); + + switch (reason) { + case ZPOOL_STATUS_MISSING_DEV_R: + case ZPOOL_STATUS_MISSING_DEV_NR: + case ZPOOL_STATUS_BAD_GUID_SUM: + (void) printf(gettext("status: One or more devices are missing " + "from the system.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_R: + case ZPOOL_STATUS_CORRUPT_LABEL_NR: + (void) printf(gettext("status: One or more devices contains " + "corrupted data.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_DATA: + (void) printf(gettext("status: The pool data is corrupted.\n")); + break; + + case ZPOOL_STATUS_OFFLINE_DEV: + (void) printf(gettext("status: One or more devices " + "are offlined.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_POOL: + (void) printf(gettext("status: The pool metadata is " + "corrupted.\n")); + break; + + case ZPOOL_STATUS_VERSION_OLDER: + (void) printf(gettext("status: The pool is formatted using an " + "older on-disk version.\n")); + break; + + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("status: The pool is formatted using an " + "incompatible version.\n")); + break; + + case ZPOOL_STATUS_HOSTID_MISMATCH: + (void) printf(gettext("status: The pool was last accessed by " + "another system.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_R: + case ZPOOL_STATUS_FAULTED_DEV_NR: + (void) printf(gettext("status: One or more devices are " + "faulted.\n")); + break; + + case ZPOOL_STATUS_BAD_LOG: + (void) printf(gettext("status: An intent log record cannot be " + "read.\n")); + break; + + default: + /* + * No other status can be seen when importing pools. + */ + assert(reason == ZPOOL_STATUS_OK); + } + + /* + * Print out an action according to the overall state of the pool. + */ + if (vs->vs_state == VDEV_STATE_HEALTHY) { + if (reason == ZPOOL_STATUS_VERSION_OLDER) + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric identifier, " + "though\n\tsome features will not be available " + "without an explicit 'zpool upgrade'.\n")); + else if (reason == ZPOOL_STATUS_HOSTID_MISMATCH) + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric " + "identifier and\n\tthe '-f' flag.\n")); + else + (void) printf(gettext("action: The pool can be " + "imported using its name or numeric " + "identifier.\n")); + } else if (vs->vs_state == VDEV_STATE_DEGRADED) { + (void) printf(gettext("action: The pool can be imported " + "despite missing or damaged devices. The\n\tfault " + "tolerance of the pool may be compromised if imported.\n")); + } else { + switch (reason) { + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("action: The pool cannot be " + "imported. Access the pool on a system running " + "newer\n\tsoftware, or recreate the pool from " + "backup.\n")); + break; + case ZPOOL_STATUS_MISSING_DEV_R: + case ZPOOL_STATUS_MISSING_DEV_NR: + case ZPOOL_STATUS_BAD_GUID_SUM: + (void) printf(gettext("action: The pool cannot be " + "imported. Attach the missing\n\tdevices and try " + "again.\n")); + break; + default: + (void) printf(gettext("action: The pool cannot be " + "imported due to damaged devices or data.\n")); + } + } + + /* + * If the state is "closed" or "can't open", and the aux state + * is "corrupt data": + */ + if (((vs->vs_state == VDEV_STATE_CLOSED) || + (vs->vs_state == VDEV_STATE_CANT_OPEN)) && + (vs->vs_aux == VDEV_AUX_CORRUPT_DATA)) { + if (pool_state == POOL_STATE_DESTROYED) + (void) printf(gettext("\tThe pool was destroyed, " + "but can be imported using the '-Df' flags.\n")); + else if (pool_state != POOL_STATE_EXPORTED) + (void) printf(gettext("\tThe pool may be active on " + "another system, but can be imported using\n\t" + "the '-f' flag.\n")); + } + + if (msgid != NULL) + (void) printf(gettext(" see: http://www.sun.com/msg/%s\n"), + msgid); + + (void) printf(gettext("config:\n\n")); + + namewidth = max_width(NULL, nvroot, 0, 0); + if (namewidth < 10) + namewidth = 10; + + print_import_config(name, nvroot, namewidth, 0, B_FALSE); + if (num_logs(nvroot) > 0) { + (void) printf(gettext("\tlogs\n")); + print_import_config(name, nvroot, namewidth, 0, B_TRUE); + } + + if (reason == ZPOOL_STATUS_BAD_GUID_SUM) { + (void) printf(gettext("\n\tAdditional devices are known to " + "be part of this pool, though their\n\texact " + "configuration cannot be determined.\n")); + } +} + +/* + * Perform the import for the given configuration. This passes the heavy + * lifting off to zpool_import_props(), and then mounts the datasets contained + * within the pool. + */ +static int +do_import(nvlist_t *config, const char *newname, const char *mntopts, + int force, nvlist_t *props, boolean_t allowfaulted) +{ + zpool_handle_t *zhp; + char *name; + uint64_t state; + uint64_t version; + int error = 0; + + verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, + &name) == 0); + + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_POOL_STATE, &state) == 0); + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_VERSION, &version) == 0); + if (version > SPA_VERSION) { + (void) fprintf(stderr, gettext("cannot import '%s': pool " + "is formatted using a newer ZFS version\n"), name); + return (1); + } else if (state != POOL_STATE_EXPORTED && !force) { + uint64_t hostid; + + if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, + &hostid) == 0) { + if ((unsigned long)hostid != gethostid()) { + char *hostname; + uint64_t timestamp; + time_t t; + + verify(nvlist_lookup_string(config, + ZPOOL_CONFIG_HOSTNAME, &hostname) == 0); + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_TIMESTAMP, ×tamp) == 0); + t = timestamp; + (void) fprintf(stderr, gettext("cannot import " + "'%s': pool may be in use from other " + "system, it was last accessed by %s " + "(hostid: 0x%lx) on %s"), name, hostname, + (unsigned long)hostid, + asctime(localtime(&t))); + (void) fprintf(stderr, gettext("use '-f' to " + "import anyway\n")); + return (1); + } + } else { + (void) fprintf(stderr, gettext("cannot import '%s': " + "pool may be in use from other system\n"), name); + (void) fprintf(stderr, gettext("use '-f' to import " + "anyway\n")); + return (1); + } + } + + if (zpool_import_props(g_zfs, config, newname, props, + allowfaulted) != 0) + return (1); + + if (newname != NULL) + name = (char *)newname; + + verify((zhp = zpool_open_canfail(g_zfs, name)) != NULL); + + if (zpool_enable_datasets(zhp, mntopts, 0) != 0) { + zpool_close(zhp); + return (1); + } + + zpool_close(zhp); + return (error); +} + +/* + * zpool import [-d dir] [-D] + * import [-o mntopts] [-o prop=value] ... [-R root] [-D] + * [-d dir | -c cachefile] [-f] -a + * import [-o mntopts] [-o prop=value] ... [-R root] [-D] + * [-d dir | -c cachefile] [-f] <pool | id> [newpool] + * + * -c Read pool information from a cachefile instead of searching + * devices. + * + * -d Scan in a specific directory, other than /dev/dsk. More than + * one directory can be specified using multiple '-d' options. + * + * -D Scan for previously destroyed pools or import all or only + * specified destroyed pools. + * + * -R Temporarily import the pool, with all mountpoints relative to + * the given root. The pool will remain exported when the machine + * is rebooted. + * + * -f Force import, even if it appears that the pool is active. + * + * -F Import even in the presence of faulted vdevs. This is an + * intentionally undocumented option for testing purposes, and + * treats the pool configuration as complete, leaving any bad + * vdevs in the FAULTED state. + * + * -a Import all pools found. + * + * -o Set property=value and/or temporary mount options (without '='). + * + * The import command scans for pools to import, and import pools based on pool + * name and GUID. The pool can also be renamed as part of the import process. + */ +int +zpool_do_import(int argc, char **argv) +{ + char **searchdirs = NULL; + int nsearch = 0; + int c; + int err; + nvlist_t *pools = NULL; + boolean_t do_all = B_FALSE; + boolean_t do_destroyed = B_FALSE; + char *mntopts = NULL; + boolean_t do_force = B_FALSE; + nvpair_t *elem; + nvlist_t *config; + uint64_t searchguid = 0; + char *searchname = NULL; + char *propval; + nvlist_t *found_config; + nvlist_t *props = NULL; + boolean_t first; + boolean_t allow_faulted = B_FALSE; + uint64_t pool_state; + char *cachefile = NULL; + + /* check options */ + while ((c = getopt(argc, argv, ":ac:d:DfFo:p:R:")) != -1) { + switch (c) { + case 'a': + do_all = B_TRUE; + break; + case 'c': + cachefile = optarg; + break; + case 'd': + if (searchdirs == NULL) { + searchdirs = safe_malloc(sizeof (char *)); + } else { + char **tmp = safe_malloc((nsearch + 1) * + sizeof (char *)); + bcopy(searchdirs, tmp, nsearch * + sizeof (char *)); + free(searchdirs); + searchdirs = tmp; + } + searchdirs[nsearch++] = optarg; + break; + case 'D': + do_destroyed = B_TRUE; + break; + case 'f': + do_force = B_TRUE; + break; + case 'F': + allow_faulted = B_TRUE; + break; + case 'o': + if ((propval = strchr(optarg, '=')) != NULL) { + *propval = '\0'; + propval++; + if (add_prop_list(optarg, propval, + &props, B_TRUE)) + goto error; + } else { + mntopts = optarg; + } + break; + case 'R': + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE)) + goto error; + if (nvlist_lookup_string(props, + zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), + &propval) == 0) + break; + if (add_prop_list(zpool_prop_to_name( + ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE)) + goto error; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (cachefile && nsearch != 0) { + (void) fprintf(stderr, gettext("-c is incompatible with -d\n")); + usage(B_FALSE); + } + + if (searchdirs == NULL) { + searchdirs = safe_malloc(sizeof (char *)); + searchdirs[0] = "/dev/dsk"; + nsearch = 1; + } + + /* check argument count */ + if (do_all) { + if (argc != 0) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + } else { + if (argc > 2) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + /* + * Check for the SYS_CONFIG privilege. We do this explicitly + * here because otherwise any attempt to discover pools will + * silently fail. + */ + if (argc == 0 && !priv_ineffect(PRIV_SYS_CONFIG)) { + (void) fprintf(stderr, gettext("cannot " + "discover pools: permission denied\n")); + free(searchdirs); + return (1); + } + } + + /* + * Depending on the arguments given, we do one of the following: + * + * <none> Iterate through all pools and display information about + * each one. + * + * -a Iterate through all pools and try to import each one. + * + * <id> Find the pool that corresponds to the given GUID/pool + * name and import that one. + * + * -D Above options applies only to destroyed pools. + */ + if (argc != 0) { + char *endptr; + + errno = 0; + searchguid = strtoull(argv[0], &endptr, 10); + if (errno != 0 || *endptr != '\0') + searchname = argv[0]; + found_config = NULL; + } + + if (cachefile) { + pools = zpool_find_import_cached(g_zfs, cachefile, searchname, + searchguid); + } else if (searchname != NULL) { + pools = zpool_find_import_byname(g_zfs, nsearch, searchdirs, + searchname); + } else { + /* + * It's OK to search by guid even if searchguid is 0. + */ + pools = zpool_find_import_byguid(g_zfs, nsearch, searchdirs, + searchguid); + } + + if (pools == NULL) { + if (argc != 0) { + (void) fprintf(stderr, gettext("cannot import '%s': " + "no such pool available\n"), argv[0]); + } + free(searchdirs); + return (1); + } + + /* + * At this point we have a list of import candidate configs. Even if + * we were searching by pool name or guid, we still need to + * post-process the list to deal with pool state and possible + * duplicate names. + */ + err = 0; + elem = NULL; + first = B_TRUE; + while ((elem = nvlist_next_nvpair(pools, elem)) != NULL) { + + verify(nvpair_value_nvlist(elem, &config) == 0); + + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, + &pool_state) == 0); + if (!do_destroyed && pool_state == POOL_STATE_DESTROYED) + continue; + if (do_destroyed && pool_state != POOL_STATE_DESTROYED) + continue; + + if (argc == 0) { + if (first) + first = B_FALSE; + else if (!do_all) + (void) printf("\n"); + + if (do_all) + err |= do_import(config, NULL, mntopts, + do_force, props, allow_faulted); + else + show_import(config); + } else if (searchname != NULL) { + char *name; + + /* + * We are searching for a pool based on name. + */ + verify(nvlist_lookup_string(config, + ZPOOL_CONFIG_POOL_NAME, &name) == 0); + + if (strcmp(name, searchname) == 0) { + if (found_config != NULL) { + (void) fprintf(stderr, gettext( + "cannot import '%s': more than " + "one matching pool\n"), searchname); + (void) fprintf(stderr, gettext( + "import by numeric ID instead\n")); + err = B_TRUE; + } + found_config = config; + } + } else { + uint64_t guid; + + /* + * Search for a pool by guid. + */ + verify(nvlist_lookup_uint64(config, + ZPOOL_CONFIG_POOL_GUID, &guid) == 0); + + if (guid == searchguid) + found_config = config; + } + } + + /* + * If we were searching for a specific pool, verify that we found a + * pool, and then do the import. + */ + if (argc != 0 && err == 0) { + if (found_config == NULL) { + (void) fprintf(stderr, gettext("cannot import '%s': " + "no such pool available\n"), argv[0]); + err = B_TRUE; + } else { + err |= do_import(found_config, argc == 1 ? NULL : + argv[1], mntopts, do_force, props, allow_faulted); + } + } + + /* + * If we were just looking for pools, report an error if none were + * found. + */ + if (argc == 0 && first) + (void) fprintf(stderr, + gettext("no pools available to import\n")); + +error: + nvlist_free(props); + nvlist_free(pools); + free(searchdirs); + + return (err ? 1 : 0); +} + +typedef struct iostat_cbdata { + zpool_list_t *cb_list; + int cb_verbose; + int cb_iteration; + int cb_namewidth; +} iostat_cbdata_t; + +static void +print_iostat_separator(iostat_cbdata_t *cb) +{ + int i = 0; + + for (i = 0; i < cb->cb_namewidth; i++) + (void) printf("-"); + (void) printf(" ----- ----- ----- ----- ----- -----\n"); +} + +static void +print_iostat_header(iostat_cbdata_t *cb) +{ + (void) printf("%*s capacity operations bandwidth\n", + cb->cb_namewidth, ""); + (void) printf("%-*s used avail read write read write\n", + cb->cb_namewidth, "pool"); + print_iostat_separator(cb); +} + +/* + * Display a single statistic. + */ +static void +print_one_stat(uint64_t value) +{ + char buf[64]; + + zfs_nicenum(value, buf, sizeof (buf)); + (void) printf(" %5s", buf); +} + +/* + * Print out all the statistics for the given vdev. This can either be the + * toplevel configuration, or called recursively. If 'name' is NULL, then this + * is a verbose output, and we don't want to display the toplevel pool stats. + */ +void +print_vdev_stats(zpool_handle_t *zhp, const char *name, nvlist_t *oldnv, + nvlist_t *newnv, iostat_cbdata_t *cb, int depth) +{ + nvlist_t **oldchild, **newchild; + uint_t c, children; + vdev_stat_t *oldvs, *newvs; + vdev_stat_t zerovs = { 0 }; + uint64_t tdelta; + double scale; + char *vname; + + if (oldnv != NULL) { + verify(nvlist_lookup_uint64_array(oldnv, ZPOOL_CONFIG_STATS, + (uint64_t **)&oldvs, &c) == 0); + } else { + oldvs = &zerovs; + } + + verify(nvlist_lookup_uint64_array(newnv, ZPOOL_CONFIG_STATS, + (uint64_t **)&newvs, &c) == 0); + + if (strlen(name) + depth > cb->cb_namewidth) + (void) printf("%*s%s", depth, "", name); + else + (void) printf("%*s%s%*s", depth, "", name, + (int)(cb->cb_namewidth - strlen(name) - depth), ""); + + tdelta = newvs->vs_timestamp - oldvs->vs_timestamp; + + if (tdelta == 0) + scale = 1.0; + else + scale = (double)NANOSEC / tdelta; + + /* only toplevel vdevs have capacity stats */ + if (newvs->vs_space == 0) { + (void) printf(" - -"); + } else { + print_one_stat(newvs->vs_alloc); + print_one_stat(newvs->vs_space - newvs->vs_alloc); + } + + print_one_stat((uint64_t)(scale * (newvs->vs_ops[ZIO_TYPE_READ] - + oldvs->vs_ops[ZIO_TYPE_READ]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_ops[ZIO_TYPE_WRITE] - + oldvs->vs_ops[ZIO_TYPE_WRITE]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_bytes[ZIO_TYPE_READ] - + oldvs->vs_bytes[ZIO_TYPE_READ]))); + + print_one_stat((uint64_t)(scale * (newvs->vs_bytes[ZIO_TYPE_WRITE] - + oldvs->vs_bytes[ZIO_TYPE_WRITE]))); + + (void) printf("\n"); + + if (!cb->cb_verbose) + return; + + if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_CHILDREN, + &newchild, &children) != 0) + return; + + if (oldnv && nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_CHILDREN, + &oldchild, &c) != 0) + return; + + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, zhp, newchild[c]); + print_vdev_stats(zhp, vname, oldnv ? oldchild[c] : NULL, + newchild[c], cb, depth + 2); + free(vname); + } + + /* + * Include level 2 ARC devices in iostat output + */ + if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_L2CACHE, + &newchild, &children) != 0) + return; + + if (oldnv && nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_L2CACHE, + &oldchild, &c) != 0) + return; + + if (children > 0) { + (void) printf("%-*s - - - - - " + "-\n", cb->cb_namewidth, "cache"); + for (c = 0; c < children; c++) { + vname = zpool_vdev_name(g_zfs, zhp, newchild[c]); + print_vdev_stats(zhp, vname, oldnv ? oldchild[c] : NULL, + newchild[c], cb, depth + 2); + free(vname); + } + } +} + +static int +refresh_iostat(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + boolean_t missing; + + /* + * If the pool has disappeared, remove it from the list and continue. + */ + if (zpool_refresh_stats(zhp, &missing) != 0) + return (-1); + + if (missing) + pool_list_remove(cb->cb_list, zhp); + + return (0); +} + +/* + * Callback to print out the iostats for the given pool. + */ +int +print_iostat(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + nvlist_t *oldconfig, *newconfig; + nvlist_t *oldnvroot, *newnvroot; + + newconfig = zpool_get_config(zhp, &oldconfig); + + if (cb->cb_iteration == 1) + oldconfig = NULL; + + verify(nvlist_lookup_nvlist(newconfig, ZPOOL_CONFIG_VDEV_TREE, + &newnvroot) == 0); + + if (oldconfig == NULL) + oldnvroot = NULL; + else + verify(nvlist_lookup_nvlist(oldconfig, ZPOOL_CONFIG_VDEV_TREE, + &oldnvroot) == 0); + + /* + * Print out the statistics for the pool. + */ + print_vdev_stats(zhp, zpool_get_name(zhp), oldnvroot, newnvroot, cb, 0); + + if (cb->cb_verbose) + print_iostat_separator(cb); + + return (0); +} + +int +get_namewidth(zpool_handle_t *zhp, void *data) +{ + iostat_cbdata_t *cb = data; + nvlist_t *config, *nvroot; + + if ((config = zpool_get_config(zhp, NULL)) != NULL) { + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if (!cb->cb_verbose) + cb->cb_namewidth = strlen(zpool_get_name(zhp)); + else + cb->cb_namewidth = max_width(zhp, nvroot, 0, 0); + } + + /* + * The width must fall into the range [10,38]. The upper limit is the + * maximum we can have and still fit in 80 columns. + */ + if (cb->cb_namewidth < 10) + cb->cb_namewidth = 10; + if (cb->cb_namewidth > 38) + cb->cb_namewidth = 38; + + return (0); +} + +/* + * zpool iostat [-v] [pool] ... [interval [count]] + * + * -v Display statistics for individual vdevs + * + * This command can be tricky because we want to be able to deal with pool + * creation/destruction as well as vdev configuration changes. The bulk of this + * processing is handled by the pool_list_* routines in zpool_iter.c. We rely + * on pool_list_update() to detect the addition of new pools. Configuration + * changes are all handled within libzfs. + */ +int +zpool_do_iostat(int argc, char **argv) +{ + int c; + int ret; + int npools; + unsigned long interval = 0, count = 0; + zpool_list_t *list; + boolean_t verbose = B_FALSE; + iostat_cbdata_t cb; + + /* check options */ + while ((c = getopt(argc, argv, "v")) != -1) { + switch (c) { + case 'v': + verbose = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* + * Determine if the last argument is an integer or a pool name + */ + if (argc > 0 && isdigit(argv[argc - 1][0])) { + char *end; + + errno = 0; + interval = strtoul(argv[argc - 1], &end, 10); + + if (*end == '\0' && errno == 0) { + if (interval == 0) { + (void) fprintf(stderr, gettext("interval " + "cannot be zero\n")); + usage(B_FALSE); + } + + /* + * Ignore the last parameter + */ + argc--; + } else { + /* + * If this is not a valid number, just plow on. The + * user will get a more informative error message later + * on. + */ + interval = 0; + } + } + + /* + * If the last argument is also an integer, then we have both a count + * and an integer. + */ + if (argc > 0 && isdigit(argv[argc - 1][0])) { + char *end; + + errno = 0; + count = interval; + interval = strtoul(argv[argc - 1], &end, 10); + + if (*end == '\0' && errno == 0) { + if (interval == 0) { + (void) fprintf(stderr, gettext("interval " + "cannot be zero\n")); + usage(B_FALSE); + } + + /* + * Ignore the last parameter + */ + argc--; + } else { + interval = 0; + } + } + + /* + * Construct the list of all interesting pools. + */ + ret = 0; + if ((list = pool_list_get(argc, argv, NULL, &ret)) == NULL) + return (1); + + if (pool_list_count(list) == 0 && argc != 0) { + pool_list_free(list); + return (1); + } + + if (pool_list_count(list) == 0 && interval == 0) { + pool_list_free(list); + (void) fprintf(stderr, gettext("no pools available\n")); + return (1); + } + + /* + * Enter the main iostat loop. + */ + cb.cb_list = list; + cb.cb_verbose = verbose; + cb.cb_iteration = 0; + cb.cb_namewidth = 0; + + for (;;) { + pool_list_update(list); + + if ((npools = pool_list_count(list)) == 0) + break; + + /* + * Refresh all statistics. This is done as an explicit step + * before calculating the maximum name width, so that any + * configuration changes are properly accounted for. + */ + (void) pool_list_iter(list, B_FALSE, refresh_iostat, &cb); + + /* + * Iterate over all pools to determine the maximum width + * for the pool / device name column across all pools. + */ + cb.cb_namewidth = 0; + (void) pool_list_iter(list, B_FALSE, get_namewidth, &cb); + + /* + * If it's the first time, or verbose mode, print the header. + */ + if (++cb.cb_iteration == 1 || verbose) + print_iostat_header(&cb); + + (void) pool_list_iter(list, B_FALSE, print_iostat, &cb); + + /* + * If there's more than one pool, and we're not in verbose mode + * (which prints a separator for us), then print a separator. + */ + if (npools > 1 && !verbose) + print_iostat_separator(&cb); + + if (verbose) + (void) printf("\n"); + + /* + * Flush the output so that redirection to a file isn't buffered + * indefinitely. + */ + (void) fflush(stdout); + + if (interval == 0) + break; + + if (count != 0 && --count == 0) + break; + + (void) sleep(interval); + } + + pool_list_free(list); + + return (ret); +} + +typedef struct list_cbdata { + boolean_t cb_scripted; + boolean_t cb_first; + zprop_list_t *cb_proplist; +} list_cbdata_t; + +/* + * Given a list of columns to display, output appropriate headers for each one. + */ +static void +print_header(zprop_list_t *pl) +{ + const char *header; + boolean_t first = B_TRUE; + boolean_t right_justify; + + for (; pl != NULL; pl = pl->pl_next) { + if (pl->pl_prop == ZPROP_INVAL) + continue; + + if (!first) + (void) printf(" "); + else + first = B_FALSE; + + header = zpool_prop_column_name(pl->pl_prop); + right_justify = zpool_prop_align_right(pl->pl_prop); + + if (pl->pl_next == NULL && !right_justify) + (void) printf("%s", header); + else if (right_justify) + (void) printf("%*s", pl->pl_width, header); + else + (void) printf("%-*s", pl->pl_width, header); + } + + (void) printf("\n"); +} + +/* + * Given a pool and a list of properties, print out all the properties according + * to the described layout. + */ +static void +print_pool(zpool_handle_t *zhp, zprop_list_t *pl, int scripted) +{ + boolean_t first = B_TRUE; + char property[ZPOOL_MAXPROPLEN]; + char *propstr; + boolean_t right_justify; + int width; + + for (; pl != NULL; pl = pl->pl_next) { + if (!first) { + if (scripted) + (void) printf("\t"); + else + (void) printf(" "); + } else { + first = B_FALSE; + } + + right_justify = B_FALSE; + if (pl->pl_prop != ZPROP_INVAL) { + if (zpool_get_prop(zhp, pl->pl_prop, property, + sizeof (property), NULL) != 0) + propstr = "-"; + else + propstr = property; + + right_justify = zpool_prop_align_right(pl->pl_prop); + } else { + propstr = "-"; + } + + width = pl->pl_width; + + /* + * If this is being called in scripted mode, or if this is the + * last column and it is left-justified, don't include a width + * format specifier. + */ + if (scripted || (pl->pl_next == NULL && !right_justify)) + (void) printf("%s", propstr); + else if (right_justify) + (void) printf("%*s", width, propstr); + else + (void) printf("%-*s", width, propstr); + } + + (void) printf("\n"); +} + +/* + * Generic callback function to list a pool. + */ +int +list_callback(zpool_handle_t *zhp, void *data) +{ + list_cbdata_t *cbp = data; + + if (cbp->cb_first) { + if (!cbp->cb_scripted) + print_header(cbp->cb_proplist); + cbp->cb_first = B_FALSE; + } + + print_pool(zhp, cbp->cb_proplist, cbp->cb_scripted); + + return (0); +} + +/* + * zpool list [-H] [-o prop[,prop]*] [pool] ... + * + * -H Scripted mode. Don't display headers, and separate properties + * by a single tab. + * -o List of properties to display. Defaults to + * "name,size,used,available,capacity,health,altroot" + * + * List all pools in the system, whether or not they're healthy. Output space + * statistics for each one, as well as health status summary. + */ +int +zpool_do_list(int argc, char **argv) +{ + int c; + int ret; + list_cbdata_t cb = { 0 }; + static char default_props[] = + "name,size,used,available,capacity,health,altroot"; + char *props = default_props; + + /* check options */ + while ((c = getopt(argc, argv, ":Ho:")) != -1) { + switch (c) { + case 'H': + cb.cb_scripted = B_TRUE; + break; + case 'o': + props = optarg; + break; + case ':': + (void) fprintf(stderr, gettext("missing argument for " + "'%c' option\n"), optopt); + usage(B_FALSE); + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (zprop_get_list(g_zfs, props, &cb.cb_proplist, ZFS_TYPE_POOL) != 0) + usage(B_FALSE); + + cb.cb_first = B_TRUE; + + ret = for_each_pool(argc, argv, B_TRUE, &cb.cb_proplist, + list_callback, &cb); + + zprop_free_list(cb.cb_proplist); + + if (argc == 0 && cb.cb_first && !cb.cb_scripted) { + (void) printf(gettext("no pools available\n")); + return (0); + } + + return (ret); +} + +static nvlist_t * +zpool_get_vdev_by_name(nvlist_t *nv, char *name) +{ + nvlist_t **child; + uint_t c, children; + nvlist_t *match; + char *path; + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + if (strncmp(name, "/dev/dsk/", 9) == 0) + name += 9; + if (strncmp(path, "/dev/dsk/", 9) == 0) + path += 9; + if (strcmp(name, path) == 0) + return (nv); + return (NULL); + } + + for (c = 0; c < children; c++) + if ((match = zpool_get_vdev_by_name(child[c], name)) != NULL) + return (match); + + return (NULL); +} + +static int +zpool_do_attach_or_replace(int argc, char **argv, int replacing) +{ + boolean_t force = B_FALSE; + int c; + nvlist_t *nvroot; + char *poolname, *old_disk, *new_disk; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + force = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if (argc < 2) { + (void) fprintf(stderr, + gettext("missing <device> specification\n")); + usage(B_FALSE); + } + + old_disk = argv[1]; + + if (argc < 3) { + if (!replacing) { + (void) fprintf(stderr, + gettext("missing <new_device> specification\n")); + usage(B_FALSE); + } + new_disk = old_disk; + argc -= 1; + argv += 1; + } else { + new_disk = argv[2]; + argc -= 2; + argv += 2; + } + + if (argc > 1) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + if (zpool_get_config(zhp, NULL) == NULL) { + (void) fprintf(stderr, gettext("pool '%s' is unavailable\n"), + poolname); + zpool_close(zhp); + return (1); + } + + nvroot = make_root_vdev(zhp, force, B_FALSE, replacing, B_FALSE, + argc, argv); + if (nvroot == NULL) { + zpool_close(zhp); + return (1); + } + + ret = zpool_vdev_attach(zhp, old_disk, new_disk, nvroot, replacing); + + nvlist_free(nvroot); + zpool_close(zhp); + + return (ret); +} + +/* + * zpool replace [-f] <pool> <device> <new_device> + * + * -f Force attach, even if <new_device> appears to be in use. + * + * Replace <device> with <new_device>. + */ +/* ARGSUSED */ +int +zpool_do_replace(int argc, char **argv) +{ + return (zpool_do_attach_or_replace(argc, argv, B_TRUE)); +} + +/* + * zpool attach [-f] <pool> <device> <new_device> + * + * -f Force attach, even if <new_device> appears to be in use. + * + * Attach <new_device> to the mirror containing <device>. If <device> is not + * part of a mirror, then <device> will be transformed into a mirror of + * <device> and <new_device>. In either case, <new_device> will begin life + * with a DTL of [0, now], and will immediately begin to resilver itself. + */ +int +zpool_do_attach(int argc, char **argv) +{ + return (zpool_do_attach_or_replace(argc, argv, B_FALSE)); +} + +/* + * zpool detach [-f] <pool> <device> + * + * -f Force detach of <device>, even if DTLs argue against it + * (not supported yet) + * + * Detach a device from a mirror. The operation will be refused if <device> + * is the last device in the mirror, or if the DTLs indicate that this device + * has the only valid copy of some data. + */ +/* ARGSUSED */ +int +zpool_do_detach(int argc, char **argv) +{ + int c; + char *poolname, *path; + zpool_handle_t *zhp; + int ret; + + /* check options */ + while ((c = getopt(argc, argv, "f")) != -1) { + switch (c) { + case 'f': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + if (argc < 2) { + (void) fprintf(stderr, + gettext("missing <device> specification\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + path = argv[1]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + ret = zpool_vdev_detach(zhp, path); + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool online <pool> <device> ... + */ +int +zpool_do_online(int argc, char **argv) +{ + int c, i; + char *poolname; + zpool_handle_t *zhp; + int ret = 0; + vdev_state_t newstate; + + /* check options */ + while ((c = getopt(argc, argv, "t")) != -1) { + switch (c) { + case 't': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device name\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_online(zhp, argv[i], 0, &newstate) == 0) { + if (newstate != VDEV_STATE_HEALTHY) { + (void) printf(gettext("warning: device '%s' " + "onlined, but remains in faulted state\n"), + argv[i]); + if (newstate == VDEV_STATE_FAULTED) + (void) printf(gettext("use 'zpool " + "clear' to restore a faulted " + "device\n")); + else + (void) printf(gettext("use 'zpool " + "replace' to replace devices " + "that are no longer present\n")); + } + } else { + ret = 1; + } + } + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool offline [-ft] <pool> <device> ... + * + * -f Force the device into the offline state, even if doing + * so would appear to compromise pool availability. + * (not supported yet) + * + * -t Only take the device off-line temporarily. The offline + * state will not be persistent across reboots. + */ +/* ARGSUSED */ +int +zpool_do_offline(int argc, char **argv) +{ + int c, i; + char *poolname; + zpool_handle_t *zhp; + int ret = 0; + boolean_t istmp = B_FALSE; + + /* check options */ + while ((c = getopt(argc, argv, "ft")) != -1) { + switch (c) { + case 't': + istmp = B_TRUE; + break; + case 'f': + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + /* get pool name and check number of arguments */ + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + if (argc < 2) { + (void) fprintf(stderr, gettext("missing device name\n")); + usage(B_FALSE); + } + + poolname = argv[0]; + + if ((zhp = zpool_open(g_zfs, poolname)) == NULL) + return (1); + + for (i = 1; i < argc; i++) { + if (zpool_vdev_offline(zhp, argv[i], istmp) != 0) + ret = 1; + } + + zpool_close(zhp); + + return (ret); +} + +/* + * zpool clear <pool> [device] + * + * Clear all errors associated with a pool or a particular device. + */ +int +zpool_do_clear(int argc, char **argv) +{ + int ret = 0; + zpool_handle_t *zhp; + char *pool, *device; + + if (argc < 2) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + + if (argc > 3) { + (void) fprintf(stderr, gettext("too many arguments\n")); + usage(B_FALSE); + } + + pool = argv[1]; + device = argc == 3 ? argv[2] : NULL; + + if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) + return (1); + + if (zpool_clear(zhp, device) != 0) + ret = 1; + + zpool_close(zhp); + + return (ret); +} + +typedef struct scrub_cbdata { + int cb_type; + int cb_argc; + char **cb_argv; +} scrub_cbdata_t; + +int +scrub_callback(zpool_handle_t *zhp, void *data) +{ + scrub_cbdata_t *cb = data; + int err; + + /* + * Ignore faulted pools. + */ + if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { + (void) fprintf(stderr, gettext("cannot scrub '%s': pool is " + "currently unavailable\n"), zpool_get_name(zhp)); + return (1); + } + + err = zpool_scrub(zhp, cb->cb_type); + + return (err != 0); +} + +/* + * zpool scrub [-s] <pool> ... + * + * -s Stop. Stops any in-progress scrub. + */ +int +zpool_do_scrub(int argc, char **argv) +{ + int c; + scrub_cbdata_t cb; + + cb.cb_type = POOL_SCRUB_EVERYTHING; + + /* check options */ + while ((c = getopt(argc, argv, "s")) != -1) { + switch (c) { + case 's': + cb.cb_type = POOL_SCRUB_NONE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + cb.cb_argc = argc; + cb.cb_argv = argv; + argc -= optind; + argv += optind; + + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + } + + return (for_each_pool(argc, argv, B_TRUE, NULL, scrub_callback, &cb)); +} + +typedef struct status_cbdata { + int cb_count; + boolean_t cb_allpools; + boolean_t cb_verbose; + boolean_t cb_explain; + boolean_t cb_first; +} status_cbdata_t; + +/* + * Print out detailed scrub status. + */ +void +print_scrub_status(nvlist_t *nvroot) +{ + vdev_stat_t *vs; + uint_t vsc; + time_t start, end, now; + double fraction_done; + uint64_t examined, total, minutes_left, minutes_taken; + char *scrub_type; + + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &vsc) == 0); + + /* + * If there's never been a scrub, there's not much to say. + */ + if (vs->vs_scrub_end == 0 && vs->vs_scrub_type == POOL_SCRUB_NONE) { + (void) printf(gettext("none requested\n")); + return; + } + + scrub_type = (vs->vs_scrub_type == POOL_SCRUB_RESILVER) ? + "resilver" : "scrub"; + + start = vs->vs_scrub_start; + end = vs->vs_scrub_end; + now = time(NULL); + examined = vs->vs_scrub_examined; + total = vs->vs_alloc; + + if (end != 0) { + minutes_taken = (uint64_t)((end - start) / 60); + + (void) printf(gettext("%s %s after %lluh%um with %llu errors " + "on %s"), + scrub_type, vs->vs_scrub_complete ? "completed" : "stopped", + (u_longlong_t)(minutes_taken / 60), + (uint_t)(minutes_taken % 60), + (u_longlong_t)vs->vs_scrub_errors, ctime(&end)); + return; + } + + if (examined == 0) + examined = 1; + if (examined > total) + total = examined; + + fraction_done = (double)examined / total; + minutes_left = (uint64_t)((now - start) * + (1 - fraction_done) / fraction_done / 60); + minutes_taken = (uint64_t)((now - start) / 60); + + (void) printf(gettext("%s in progress for %lluh%um, %.2f%% done, " + "%lluh%um to go\n"), + scrub_type, (u_longlong_t)(minutes_taken / 60), + (uint_t)(minutes_taken % 60), 100 * fraction_done, + (u_longlong_t)(minutes_left / 60), (uint_t)(minutes_left % 60)); +} + +typedef struct spare_cbdata { + uint64_t cb_guid; + zpool_handle_t *cb_zhp; +} spare_cbdata_t; + +static boolean_t +find_vdev(nvlist_t *nv, uint64_t search) +{ + uint64_t guid; + nvlist_t **child; + uint_t c, children; + + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0 && + search == guid) + return (B_TRUE); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) == 0) { + for (c = 0; c < children; c++) + if (find_vdev(child[c], search)) + return (B_TRUE); + } + + return (B_FALSE); +} + +static int +find_spare(zpool_handle_t *zhp, void *data) +{ + spare_cbdata_t *cbp = data; + nvlist_t *config, *nvroot; + + config = zpool_get_config(zhp, NULL); + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + + if (find_vdev(nvroot, cbp->cb_guid)) { + cbp->cb_zhp = zhp; + return (1); + } + + zpool_close(zhp); + return (0); +} + +/* + * Print out configuration state as requested by status_callback. + */ +void +print_status_config(zpool_handle_t *zhp, const char *name, nvlist_t *nv, + int namewidth, int depth, boolean_t isspare, boolean_t print_logs) +{ + nvlist_t **child; + uint_t c, children; + vdev_stat_t *vs; + char rbuf[6], wbuf[6], cbuf[6], repaired[7]; + char *vname; + uint64_t notpresent; + spare_cbdata_t cb; + char *state; + + verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + children = 0; + + state = zpool_state_to_name(vs->vs_state, vs->vs_aux); + if (isspare) { + /* + * For hot spares, we use the terms 'INUSE' and 'AVAILABLE' for + * online drives. + */ + if (vs->vs_aux == VDEV_AUX_SPARED) + state = "INUSE"; + else if (vs->vs_state == VDEV_STATE_HEALTHY) + state = "AVAIL"; + } + + (void) printf("\t%*s%-*s %-8s", depth, "", namewidth - depth, + name, state); + + if (!isspare) { + zfs_nicenum(vs->vs_read_errors, rbuf, sizeof (rbuf)); + zfs_nicenum(vs->vs_write_errors, wbuf, sizeof (wbuf)); + zfs_nicenum(vs->vs_checksum_errors, cbuf, sizeof (cbuf)); + (void) printf(" %5s %5s %5s", rbuf, wbuf, cbuf); + } + + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, + ¬present) == 0) { + char *path; + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + (void) printf(" was %s", path); + } else if (vs->vs_aux != 0) { + (void) printf(" "); + + switch (vs->vs_aux) { + case VDEV_AUX_OPEN_FAILED: + (void) printf(gettext("cannot open")); + break; + + case VDEV_AUX_BAD_GUID_SUM: + (void) printf(gettext("missing device")); + break; + + case VDEV_AUX_NO_REPLICAS: + (void) printf(gettext("insufficient replicas")); + break; + + case VDEV_AUX_VERSION_NEWER: + (void) printf(gettext("newer version")); + break; + + case VDEV_AUX_SPARED: + verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, + &cb.cb_guid) == 0); + if (zpool_iter(g_zfs, find_spare, &cb) == 1) { + if (strcmp(zpool_get_name(cb.cb_zhp), + zpool_get_name(zhp)) == 0) + (void) printf(gettext("currently in " + "use")); + else + (void) printf(gettext("in use by " + "pool '%s'"), + zpool_get_name(cb.cb_zhp)); + zpool_close(cb.cb_zhp); + } else { + (void) printf(gettext("currently in use")); + } + break; + + case VDEV_AUX_ERR_EXCEEDED: + (void) printf(gettext("too many errors")); + break; + + case VDEV_AUX_IO_FAILURE: + (void) printf(gettext("experienced I/O failures")); + break; + + case VDEV_AUX_BAD_LOG: + (void) printf(gettext("bad intent log")); + break; + + default: + (void) printf(gettext("corrupted data")); + break; + } + } else if (vs->vs_scrub_repaired != 0 && children == 0) { + /* + * Report bytes resilvered/repaired on leaf devices. + */ + zfs_nicenum(vs->vs_scrub_repaired, repaired, sizeof (repaired)); + (void) printf(gettext(" %s %s"), repaired, + (vs->vs_scrub_type == POOL_SCRUB_RESILVER) ? + "resilvered" : "repaired"); + } + + (void) printf("\n"); + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if ((is_log && !print_logs) || (!is_log && print_logs)) + continue; + vname = zpool_vdev_name(g_zfs, zhp, child[c]); + print_status_config(zhp, vname, child[c], + namewidth, depth + 2, isspare, B_FALSE); + free(vname); + } +} + +static void +print_error_log(zpool_handle_t *zhp) +{ + nvlist_t *nverrlist = NULL; + nvpair_t *elem; + char *pathname; + size_t len = MAXPATHLEN * 2; + + if (zpool_get_errlog(zhp, &nverrlist) != 0) { + (void) printf("errors: List of errors unavailable " + "(insufficient privileges)\n"); + return; + } + + (void) printf("errors: Permanent errors have been " + "detected in the following files:\n\n"); + + pathname = safe_malloc(len); + elem = NULL; + while ((elem = nvlist_next_nvpair(nverrlist, elem)) != NULL) { + nvlist_t *nv; + uint64_t dsobj, obj; + + verify(nvpair_value_nvlist(elem, &nv) == 0); + verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_DATASET, + &dsobj) == 0); + verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_OBJECT, + &obj) == 0); + zpool_obj_to_path(zhp, dsobj, obj, pathname, len); + (void) printf("%7s %s\n", "", pathname); + } + free(pathname); + nvlist_free(nverrlist); +} + +static void +print_spares(zpool_handle_t *zhp, nvlist_t **spares, uint_t nspares, + int namewidth) +{ + uint_t i; + char *name; + + if (nspares == 0) + return; + + (void) printf(gettext("\tspares\n")); + + for (i = 0; i < nspares; i++) { + name = zpool_vdev_name(g_zfs, zhp, spares[i]); + print_status_config(zhp, name, spares[i], + namewidth, 2, B_TRUE, B_FALSE); + free(name); + } +} + +static void +print_l2cache(zpool_handle_t *zhp, nvlist_t **l2cache, uint_t nl2cache, + int namewidth) +{ + uint_t i; + char *name; + + if (nl2cache == 0) + return; + + (void) printf(gettext("\tcache\n")); + + for (i = 0; i < nl2cache; i++) { + name = zpool_vdev_name(g_zfs, zhp, l2cache[i]); + print_status_config(zhp, name, l2cache[i], + namewidth, 2, B_FALSE, B_FALSE); + free(name); + } +} + +/* + * Display a summary of pool status. Displays a summary such as: + * + * pool: tank + * status: DEGRADED + * reason: One or more devices ... + * see: http://www.sun.com/msg/ZFS-xxxx-01 + * config: + * mirror DEGRADED + * c1t0d0 OK + * c2t0d0 UNAVAIL + * + * When given the '-v' option, we print out the complete config. If the '-e' + * option is specified, then we print out error rate information as well. + */ +int +status_callback(zpool_handle_t *zhp, void *data) +{ + status_cbdata_t *cbp = data; + nvlist_t *config, *nvroot; + char *msgid; + int reason; + const char *health; + uint_t c; + vdev_stat_t *vs; + + config = zpool_get_config(zhp, NULL); + reason = zpool_get_status(zhp, &msgid); + + cbp->cb_count++; + + /* + * If we were given 'zpool status -x', only report those pools with + * problems. + */ + if (reason == ZPOOL_STATUS_OK && cbp->cb_explain) { + if (!cbp->cb_allpools) { + (void) printf(gettext("pool '%s' is healthy\n"), + zpool_get_name(zhp)); + if (cbp->cb_first) + cbp->cb_first = B_FALSE; + } + return (0); + } + + if (cbp->cb_first) + cbp->cb_first = B_FALSE; + else + (void) printf("\n"); + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS, + (uint64_t **)&vs, &c) == 0); + health = zpool_state_to_name(vs->vs_state, vs->vs_aux); + + (void) printf(gettext(" pool: %s\n"), zpool_get_name(zhp)); + (void) printf(gettext(" state: %s\n"), health); + + switch (reason) { + case ZPOOL_STATUS_MISSING_DEV_R: + (void) printf(gettext("status: One or more devices could not " + "be opened. Sufficient replicas exist for\n\tthe pool to " + "continue functioning in a degraded state.\n")); + (void) printf(gettext("action: Attach the missing device and " + "online it using 'zpool online'.\n")); + break; + + case ZPOOL_STATUS_MISSING_DEV_NR: + (void) printf(gettext("status: One or more devices could not " + "be opened. There are insufficient\n\treplicas for the " + "pool to continue functioning.\n")); + (void) printf(gettext("action: Attach the missing device and " + "online it using 'zpool online'.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_R: + (void) printf(gettext("status: One or more devices could not " + "be used because the label is missing or\n\tinvalid. " + "Sufficient replicas exist for the pool to continue\n\t" + "functioning in a degraded state.\n")); + (void) printf(gettext("action: Replace the device using " + "'zpool replace'.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_LABEL_NR: + (void) printf(gettext("status: One or more devices could not " + "be used because the label is missing \n\tor invalid. " + "There are insufficient replicas for the pool to " + "continue\n\tfunctioning.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source.\n")); + break; + + case ZPOOL_STATUS_FAILING_DEV: + (void) printf(gettext("status: One or more devices has " + "experienced an unrecoverable error. An\n\tattempt was " + "made to correct the error. Applications are " + "unaffected.\n")); + (void) printf(gettext("action: Determine if the device needs " + "to be replaced, and clear the errors\n\tusing " + "'zpool clear' or replace the device with 'zpool " + "replace'.\n")); + break; + + case ZPOOL_STATUS_OFFLINE_DEV: + (void) printf(gettext("status: One or more devices has " + "been taken offline by the administrator.\n\tSufficient " + "replicas exist for the pool to continue functioning in " + "a\n\tdegraded state.\n")); + (void) printf(gettext("action: Online the device using " + "'zpool online' or replace the device with\n\t'zpool " + "replace'.\n")); + break; + + case ZPOOL_STATUS_RESILVERING: + (void) printf(gettext("status: One or more devices is " + "currently being resilvered. The pool will\n\tcontinue " + "to function, possibly in a degraded state.\n")); + (void) printf(gettext("action: Wait for the resilver to " + "complete.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_DATA: + (void) printf(gettext("status: One or more devices has " + "experienced an error resulting in data\n\tcorruption. " + "Applications may be affected.\n")); + (void) printf(gettext("action: Restore the file in question " + "if possible. Otherwise restore the\n\tentire pool from " + "backup.\n")); + break; + + case ZPOOL_STATUS_CORRUPT_POOL: + (void) printf(gettext("status: The pool metadata is corrupted " + "and the pool cannot be opened.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source.\n")); + break; + + case ZPOOL_STATUS_VERSION_OLDER: + (void) printf(gettext("status: The pool is formatted using an " + "older on-disk format. The pool can\n\tstill be used, but " + "some features are unavailable.\n")); + (void) printf(gettext("action: Upgrade the pool using 'zpool " + "upgrade'. Once this is done, the\n\tpool will no longer " + "be accessible on older software versions.\n")); + break; + + case ZPOOL_STATUS_VERSION_NEWER: + (void) printf(gettext("status: The pool has been upgraded to a " + "newer, incompatible on-disk version.\n\tThe pool cannot " + "be accessed on this system.\n")); + (void) printf(gettext("action: Access the pool from a system " + "running more recent software, or\n\trestore the pool from " + "backup.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_R: + (void) printf(gettext("status: One or more devices are " + "faulted in response to persistent errors.\n\tSufficient " + "replicas exist for the pool to continue functioning " + "in a\n\tdegraded state.\n")); + (void) printf(gettext("action: Replace the faulted device, " + "or use 'zpool clear' to mark the device\n\trepaired.\n")); + break; + + case ZPOOL_STATUS_FAULTED_DEV_NR: + (void) printf(gettext("status: One or more devices are " + "faulted in response to persistent errors. There are " + "insufficient replicas for the pool to\n\tcontinue " + "functioning.\n")); + (void) printf(gettext("action: Destroy and re-create the pool " + "from a backup source. Manually marking the device\n" + "\trepaired using 'zpool clear' may allow some data " + "to be recovered.\n")); + break; + + case ZPOOL_STATUS_IO_FAILURE_WAIT: + case ZPOOL_STATUS_IO_FAILURE_CONTINUE: + (void) printf(gettext("status: One or more devices are " + "faulted in response to IO failures.\n")); + (void) printf(gettext("action: Make sure the affected devices " + "are connected, then run 'zpool clear'.\n")); + break; + + case ZPOOL_STATUS_BAD_LOG: + (void) printf(gettext("status: An intent log record " + "could not be read.\n" + "\tWaiting for adminstrator intervention to fix the " + "faulted pool.\n")); + (void) printf(gettext("action: Either restore the affected " + "device(s) and run 'zpool online',\n" + "\tor ignore the intent log records by running " + "'zpool clear'.\n")); + break; + + default: + /* + * The remaining errors can't actually be generated, yet. + */ + assert(reason == ZPOOL_STATUS_OK); + } + + if (msgid != NULL) + (void) printf(gettext(" see: http://www.sun.com/msg/%s\n"), + msgid); + + if (config != NULL) { + int namewidth; + uint64_t nerr; + nvlist_t **spares, **l2cache; + uint_t nspares, nl2cache; + + + (void) printf(gettext(" scrub: ")); + print_scrub_status(nvroot); + + namewidth = max_width(zhp, nvroot, 0, 0); + if (namewidth < 10) + namewidth = 10; + + (void) printf(gettext("config:\n\n")); + (void) printf(gettext("\t%-*s %-8s %5s %5s %5s\n"), namewidth, + "NAME", "STATE", "READ", "WRITE", "CKSUM"); + print_status_config(zhp, zpool_get_name(zhp), nvroot, + namewidth, 0, B_FALSE, B_FALSE); + if (num_logs(nvroot) > 0) + print_status_config(zhp, "logs", nvroot, namewidth, 0, + B_FALSE, B_TRUE); + + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, + &l2cache, &nl2cache) == 0) + print_l2cache(zhp, l2cache, nl2cache, namewidth); + + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + &spares, &nspares) == 0) + print_spares(zhp, spares, nspares, namewidth); + + if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT, + &nerr) == 0) { + nvlist_t *nverrlist = NULL; + + /* + * If the approximate error count is small, get a + * precise count by fetching the entire log and + * uniquifying the results. + */ + if (nerr > 0 && nerr < 100 && !cbp->cb_verbose && + zpool_get_errlog(zhp, &nverrlist) == 0) { + nvpair_t *elem; + + elem = NULL; + nerr = 0; + while ((elem = nvlist_next_nvpair(nverrlist, + elem)) != NULL) { + nerr++; + } + } + nvlist_free(nverrlist); + + (void) printf("\n"); + + if (nerr == 0) + (void) printf(gettext("errors: No known data " + "errors\n")); + else if (!cbp->cb_verbose) + (void) printf(gettext("errors: %llu data " + "errors, use '-v' for a list\n"), + (u_longlong_t)nerr); + else + print_error_log(zhp); + } + } else { + (void) printf(gettext("config: The configuration cannot be " + "determined.\n")); + } + + return (0); +} + +/* + * zpool status [-vx] [pool] ... + * + * -v Display complete error logs + * -x Display only pools with potential problems + * + * Describes the health status of all pools or some subset. + */ +int +zpool_do_status(int argc, char **argv) +{ + int c; + int ret; + status_cbdata_t cb = { 0 }; + + /* check options */ + while ((c = getopt(argc, argv, "vx")) != -1) { + switch (c) { + case 'v': + cb.cb_verbose = B_TRUE; + break; + case 'x': + cb.cb_explain = B_TRUE; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + cb.cb_first = B_TRUE; + + if (argc == 0) + cb.cb_allpools = B_TRUE; + + ret = for_each_pool(argc, argv, B_TRUE, NULL, status_callback, &cb); + + if (argc == 0 && cb.cb_count == 0) + (void) printf(gettext("no pools available\n")); + else if (cb.cb_explain && cb.cb_first && cb.cb_allpools) + (void) printf(gettext("all pools are healthy\n")); + + return (ret); +} + +typedef struct upgrade_cbdata { + int cb_all; + int cb_first; + int cb_newer; + int cb_argc; + uint64_t cb_version; + char **cb_argv; +} upgrade_cbdata_t; + +static int +upgrade_cb(zpool_handle_t *zhp, void *arg) +{ + upgrade_cbdata_t *cbp = arg; + nvlist_t *config; + uint64_t version; + int ret = 0; + + config = zpool_get_config(zhp, NULL); + verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, + &version) == 0); + + if (!cbp->cb_newer && version < SPA_VERSION) { + if (!cbp->cb_all) { + if (cbp->cb_first) { + (void) printf(gettext("The following pools are " + "out of date, and can be upgraded. After " + "being\nupgraded, these pools will no " + "longer be accessible by older software " + "versions.\n\n")); + (void) printf(gettext("VER POOL\n")); + (void) printf(gettext("--- ------------\n")); + cbp->cb_first = B_FALSE; + } + + (void) printf("%2llu %s\n", (u_longlong_t)version, + zpool_get_name(zhp)); + } else { + cbp->cb_first = B_FALSE; + ret = zpool_upgrade(zhp, cbp->cb_version); + if (!ret) { + (void) printf(gettext("Successfully upgraded " + "'%s'\n\n"), zpool_get_name(zhp)); + } + } + } else if (cbp->cb_newer && version > SPA_VERSION) { + assert(!cbp->cb_all); + + if (cbp->cb_first) { + (void) printf(gettext("The following pools are " + "formatted using a newer software version and\n" + "cannot be accessed on the current system.\n\n")); + (void) printf(gettext("VER POOL\n")); + (void) printf(gettext("--- ------------\n")); + cbp->cb_first = B_FALSE; + } + + (void) printf("%2llu %s\n", (u_longlong_t)version, + zpool_get_name(zhp)); + } + + zpool_close(zhp); + return (ret); +} + +/* ARGSUSED */ +static int +upgrade_one(zpool_handle_t *zhp, void *data) +{ + upgrade_cbdata_t *cbp = data; + uint64_t cur_version; + int ret; + + if (strcmp("log", zpool_get_name(zhp)) == 0) { + (void) printf(gettext("'log' is now a reserved word\n" + "Pool 'log' must be renamed using export and import" + " to upgrade.\n")); + return (1); + } + + cur_version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); + if (cur_version > cbp->cb_version) { + (void) printf(gettext("Pool '%s' is already formatted " + "using more current version '%llu'.\n"), + zpool_get_name(zhp), cur_version); + return (0); + } + if (cur_version == cbp->cb_version) { + (void) printf(gettext("Pool '%s' is already formatted " + "using the current version.\n"), zpool_get_name(zhp)); + return (0); + } + + ret = zpool_upgrade(zhp, cbp->cb_version); + + if (!ret) { + (void) printf(gettext("Successfully upgraded '%s' " + "from version %llu to version %llu\n\n"), + zpool_get_name(zhp), (u_longlong_t)cur_version, + (u_longlong_t)cbp->cb_version); + } + + return (ret != 0); +} + +/* + * zpool upgrade + * zpool upgrade -v + * zpool upgrade [-V version] <-a | pool ...> + * + * With no arguments, display downrev'd ZFS pool available for upgrade. + * Individual pools can be upgraded by specifying the pool, and '-a' will + * upgrade all pools. + */ +int +zpool_do_upgrade(int argc, char **argv) +{ + int c; + upgrade_cbdata_t cb = { 0 }; + int ret = 0; + boolean_t showversions = B_FALSE; + char *end; + + + /* check options */ + while ((c = getopt(argc, argv, "avV:")) != -1) { + switch (c) { + case 'a': + cb.cb_all = B_TRUE; + break; + case 'v': + showversions = B_TRUE; + break; + case 'V': + cb.cb_version = strtoll(optarg, &end, 10); + if (*end != '\0' || cb.cb_version > SPA_VERSION || + cb.cb_version < SPA_VERSION_1) { + (void) fprintf(stderr, + gettext("invalid version '%s'\n"), optarg); + usage(B_FALSE); + } + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + + cb.cb_argc = argc; + cb.cb_argv = argv; + argc -= optind; + argv += optind; + + if (cb.cb_version == 0) { + cb.cb_version = SPA_VERSION; + } else if (!cb.cb_all && argc == 0) { + (void) fprintf(stderr, gettext("-V option is " + "incompatible with other arguments\n")); + usage(B_FALSE); + } + + if (showversions) { + if (cb.cb_all || argc != 0) { + (void) fprintf(stderr, gettext("-v option is " + "incompatible with other arguments\n")); + usage(B_FALSE); + } + } else if (cb.cb_all) { + if (argc != 0) { + (void) fprintf(stderr, gettext("-a option should not " + "be used along with a pool name\n")); + usage(B_FALSE); + } + } + + (void) printf(gettext("This system is currently running " + "ZFS pool version %llu.\n\n"), SPA_VERSION); + cb.cb_first = B_TRUE; + if (showversions) { + (void) printf(gettext("The following versions are " + "supported:\n\n")); + (void) printf(gettext("VER DESCRIPTION\n")); + (void) printf("--- -----------------------------------------" + "---------------\n"); + (void) printf(gettext(" 1 Initial ZFS version\n")); + (void) printf(gettext(" 2 Ditto blocks " + "(replicated metadata)\n")); + (void) printf(gettext(" 3 Hot spares and double parity " + "RAID-Z\n")); + (void) printf(gettext(" 4 zpool history\n")); + (void) printf(gettext(" 5 Compression using the gzip " + "algorithm\n")); + (void) printf(gettext(" 6 bootfs pool property\n")); + (void) printf(gettext(" 7 Separate intent log devices\n")); + (void) printf(gettext(" 8 Delegated administration\n")); + (void) printf(gettext(" 9 refquota and refreservation " + "properties\n")); + (void) printf(gettext(" 10 Cache devices\n")); + (void) printf(gettext(" 11 Improved scrub performance\n")); + (void) printf(gettext(" 12 Snapshot properties\n")); + (void) printf(gettext(" 13 snapused property\n")); + (void) printf(gettext(" 14 passthrough-x aclinherit " + "support\n")); + (void) printf(gettext("For more information on a particular " + "version, including supported releases, see:\n\n")); + (void) printf("http://www.opensolaris.org/os/community/zfs/" + "version/N\n\n"); + (void) printf(gettext("Where 'N' is the version number.\n")); + } else if (argc == 0) { + int notfound; + + ret = zpool_iter(g_zfs, upgrade_cb, &cb); + notfound = cb.cb_first; + + if (!cb.cb_all && ret == 0) { + if (!cb.cb_first) + (void) printf("\n"); + cb.cb_first = B_TRUE; + cb.cb_newer = B_TRUE; + ret = zpool_iter(g_zfs, upgrade_cb, &cb); + if (!cb.cb_first) { + notfound = B_FALSE; + (void) printf("\n"); + } + } + + if (ret == 0) { + if (notfound) + (void) printf(gettext("All pools are formatted " + "using this version.\n")); + else if (!cb.cb_all) + (void) printf(gettext("Use 'zpool upgrade -v' " + "for a list of available versions and " + "their associated\nfeatures.\n")); + } + } else { + ret = for_each_pool(argc, argv, B_FALSE, NULL, + upgrade_one, &cb); + } + + return (ret); +} + +typedef struct hist_cbdata { + boolean_t first; + int longfmt; + int internal; +} hist_cbdata_t; + +char *hist_event_table[LOG_END] = { + "invalid event", + "pool create", + "vdev add", + "pool remove", + "pool destroy", + "pool export", + "pool import", + "vdev attach", + "vdev replace", + "vdev detach", + "vdev online", + "vdev offline", + "vdev upgrade", + "pool clear", + "pool scrub", + "pool property set", + "create", + "clone", + "destroy", + "destroy_begin_sync", + "inherit", + "property set", + "quota set", + "permission update", + "permission remove", + "permission who remove", + "promote", + "receive", + "rename", + "reservation set", + "replay_inc_sync", + "replay_full_sync", + "rollback", + "snapshot", + "filesystem version upgrade", + "refquota set", + "refreservation set", + "pool scrub done", +}; + +/* + * Print out the command history for a specific pool. + */ +static int +get_history_one(zpool_handle_t *zhp, void *data) +{ + nvlist_t *nvhis; + nvlist_t **records; + uint_t numrecords; + char *cmdstr; + char *pathstr; + uint64_t dst_time; + time_t tsec; + struct tm t; + char tbuf[30]; + int ret, i; + uint64_t who; + struct passwd *pwd; + char *hostname; + char *zonename; + char internalstr[MAXPATHLEN]; + hist_cbdata_t *cb = (hist_cbdata_t *)data; + uint64_t txg; + uint64_t ievent; + + cb->first = B_FALSE; + + (void) printf(gettext("History for '%s':\n"), zpool_get_name(zhp)); + + if ((ret = zpool_get_history(zhp, &nvhis)) != 0) + return (ret); + + verify(nvlist_lookup_nvlist_array(nvhis, ZPOOL_HIST_RECORD, + &records, &numrecords) == 0); + for (i = 0; i < numrecords; i++) { + if (nvlist_lookup_uint64(records[i], ZPOOL_HIST_TIME, + &dst_time) != 0) + continue; + + /* is it an internal event or a standard event? */ + if (nvlist_lookup_string(records[i], ZPOOL_HIST_CMD, + &cmdstr) != 0) { + if (cb->internal == 0) + continue; + + if (nvlist_lookup_uint64(records[i], + ZPOOL_HIST_INT_EVENT, &ievent) != 0) + continue; + verify(nvlist_lookup_uint64(records[i], + ZPOOL_HIST_TXG, &txg) == 0); + verify(nvlist_lookup_string(records[i], + ZPOOL_HIST_INT_STR, &pathstr) == 0); + if (ievent >= LOG_END) + continue; + (void) snprintf(internalstr, + sizeof (internalstr), + "[internal %s txg:%lld] %s", + hist_event_table[ievent], txg, + pathstr); + cmdstr = internalstr; + } + tsec = dst_time; + (void) localtime_r(&tsec, &t); + (void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t); + (void) printf("%s %s", tbuf, cmdstr); + + if (!cb->longfmt) { + (void) printf("\n"); + continue; + } + (void) printf(" ["); + if (nvlist_lookup_uint64(records[i], + ZPOOL_HIST_WHO, &who) == 0) { + pwd = getpwuid((uid_t)who); + if (pwd) + (void) printf("user %s on", + pwd->pw_name); + else + (void) printf("user %d on", + (int)who); + } else { + (void) printf(gettext("no info]\n")); + continue; + } + if (nvlist_lookup_string(records[i], + ZPOOL_HIST_HOST, &hostname) == 0) { + (void) printf(" %s", hostname); + } + if (nvlist_lookup_string(records[i], + ZPOOL_HIST_ZONE, &zonename) == 0) { + (void) printf(":%s", zonename); + } + + (void) printf("]"); + (void) printf("\n"); + } + (void) printf("\n"); + nvlist_free(nvhis); + + return (ret); +} + +/* + * zpool history <pool> + * + * Displays the history of commands that modified pools. + */ + + +int +zpool_do_history(int argc, char **argv) +{ + hist_cbdata_t cbdata = { 0 }; + int ret; + int c; + + cbdata.first = B_TRUE; + /* check options */ + while ((c = getopt(argc, argv, "li")) != -1) { + switch (c) { + case 'l': + cbdata.longfmt = 1; + break; + case 'i': + cbdata.internal = 1; + break; + case '?': + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + optopt); + usage(B_FALSE); + } + } + argc -= optind; + argv += optind; + + ret = for_each_pool(argc, argv, B_FALSE, NULL, get_history_one, + &cbdata); + + if (argc == 0 && cbdata.first == B_TRUE) { + (void) printf(gettext("no pools available\n")); + return (0); + } + + return (ret); +} + +static int +get_callback(zpool_handle_t *zhp, void *data) +{ + zprop_get_cbdata_t *cbp = (zprop_get_cbdata_t *)data; + char value[MAXNAMELEN]; + zprop_source_t srctype; + zprop_list_t *pl; + + for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) { + + /* + * Skip the special fake placeholder. This will also skip + * over the name property when 'all' is specified. + */ + if (pl->pl_prop == ZPOOL_PROP_NAME && + pl == cbp->cb_proplist) + continue; + + if (zpool_get_prop(zhp, pl->pl_prop, + value, sizeof (value), &srctype) != 0) + continue; + + zprop_print_one_property(zpool_get_name(zhp), cbp, + zpool_prop_to_name(pl->pl_prop), value, srctype, NULL); + } + return (0); +} + +int +zpool_do_get(int argc, char **argv) +{ + zprop_get_cbdata_t cb = { 0 }; + zprop_list_t fake_name = { 0 }; + int ret; + + if (argc < 3) + usage(B_FALSE); + + cb.cb_first = B_TRUE; + cb.cb_sources = ZPROP_SRC_ALL; + cb.cb_columns[0] = GET_COL_NAME; + cb.cb_columns[1] = GET_COL_PROPERTY; + cb.cb_columns[2] = GET_COL_VALUE; + cb.cb_columns[3] = GET_COL_SOURCE; + cb.cb_type = ZFS_TYPE_POOL; + + if (zprop_get_list(g_zfs, argv[1], &cb.cb_proplist, + ZFS_TYPE_POOL) != 0) + usage(B_FALSE); + + if (cb.cb_proplist != NULL) { + fake_name.pl_prop = ZPOOL_PROP_NAME; + fake_name.pl_width = strlen(gettext("NAME")); + fake_name.pl_next = cb.cb_proplist; + cb.cb_proplist = &fake_name; + } + + ret = for_each_pool(argc - 2, argv + 2, B_TRUE, &cb.cb_proplist, + get_callback, &cb); + + if (cb.cb_proplist == &fake_name) + zprop_free_list(fake_name.pl_next); + else + zprop_free_list(cb.cb_proplist); + + return (ret); +} + +typedef struct set_cbdata { + char *cb_propname; + char *cb_value; + boolean_t cb_any_successful; +} set_cbdata_t; + +int +set_callback(zpool_handle_t *zhp, void *data) +{ + int error; + set_cbdata_t *cb = (set_cbdata_t *)data; + + error = zpool_set_prop(zhp, cb->cb_propname, cb->cb_value); + + if (!error) + cb->cb_any_successful = B_TRUE; + + return (error); +} + +int +zpool_do_set(int argc, char **argv) +{ + set_cbdata_t cb = { 0 }; + int error; + + if (argc > 1 && argv[1][0] == '-') { + (void) fprintf(stderr, gettext("invalid option '%c'\n"), + argv[1][1]); + usage(B_FALSE); + } + + if (argc < 2) { + (void) fprintf(stderr, gettext("missing property=value " + "argument\n")); + usage(B_FALSE); + } + + if (argc < 3) { + (void) fprintf(stderr, gettext("missing pool name\n")); + usage(B_FALSE); + } + + if (argc > 3) { + (void) fprintf(stderr, gettext("too many pool names\n")); + usage(B_FALSE); + } + + cb.cb_propname = argv[1]; + cb.cb_value = strchr(cb.cb_propname, '='); + if (cb.cb_value == NULL) { + (void) fprintf(stderr, gettext("missing value in " + "property=value argument\n")); + usage(B_FALSE); + } + + *(cb.cb_value) = '\0'; + cb.cb_value++; + + error = for_each_pool(argc - 2, argv + 2, B_TRUE, NULL, + set_callback, &cb); + + return (error); +} + +static int +find_command_idx(char *command, int *idx) +{ + int i; + + for (i = 0; i < NCOMMAND; i++) { + if (command_table[i].name == NULL) + continue; + + if (strcmp(command, command_table[i].name) == 0) { + *idx = i; + return (0); + } + } + return (1); +} + +int +main(int argc, char **argv) +{ + int ret; + int i; + char *cmdname; + + (void) setlocale(LC_ALL, ""); + (void) textdomain(TEXT_DOMAIN); + + if ((g_zfs = libzfs_init()) == NULL) { + (void) fprintf(stderr, gettext("internal error: failed to " + "initialize ZFS library\n")); + return (1); + } + + libzfs_print_on_error(g_zfs, B_TRUE); + + opterr = 0; + + /* + * Make sure the user has specified some command. + */ + if (argc < 2) { + (void) fprintf(stderr, gettext("missing command\n")); + usage(B_FALSE); + } + + cmdname = argv[1]; + + /* + * Special case '-?' + */ + if (strcmp(cmdname, "-?") == 0) + usage(B_TRUE); + + zpool_set_history_str("zpool", argc, argv, history_str); + verify(zpool_stage_history(g_zfs, history_str) == 0); + + /* + * Run the appropriate command. + */ + if (find_command_idx(cmdname, &i) == 0) { + current_command = &command_table[i]; + ret = command_table[i].func(argc - 1, argv + 1); + } else if (strchr(cmdname, '=')) { + verify(find_command_idx("set", &i) == 0); + current_command = &command_table[i]; + ret = command_table[i].func(argc, argv); + } else if (strcmp(cmdname, "freeze") == 0 && argc == 3) { + /* + * 'freeze' is a vile debugging abomination, so we treat + * it as such. + */ + char buf[16384]; + int fd = open(ZFS_DEV, O_RDWR); + (void) strcpy((void *)buf, argv[2]); + return (!!ioctl(fd, ZFS_IOC_POOL_FREEZE, buf)); + } else { + (void) fprintf(stderr, gettext("unrecognized " + "command '%s'\n"), cmdname); + usage(B_FALSE); + } + + libzfs_fini(g_zfs); + + /* + * The 'ZFS_ABORT' environment variable causes us to dump core on exit + * for the purposes of running ::findleaks. + */ + if (getenv("ZFS_ABORT") != NULL) { + (void) printf("dumping core by request\n"); + abort(); + } + + return (ret); +} diff --git a/cmd/zpool/zpool_util.c b/cmd/zpool/zpool_util.c new file mode 100644 index 000000000..f44da4ff6 --- /dev/null +++ b/cmd/zpool/zpool_util.c @@ -0,0 +1,104 @@ +/* + * 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 http://www.opensolaris.org/os/licensing. + * 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 2007 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#pragma ident "%Z%%M% %I% %E% SMI" + +#include <errno.h> +#include <libgen.h> +#include <libintl.h> +#include <stdio.h> +#include <stdlib.h> +#include <strings.h> + +#include "zpool_util.h" + +/* + * Utility function to guarantee malloc() success. + */ +void * +safe_malloc(size_t size) +{ + void *data; + + if ((data = calloc(1, size)) == NULL) { + (void) fprintf(stderr, "internal error: out of memory\n"); + exit(1); + } + + return (data); +} + +/* + * Same as above, but for strdup() + */ +char * +safe_strdup(const char *str) +{ + char *ret; + + if ((ret = strdup(str)) == NULL) { + (void) fprintf(stderr, "internal error: out of memory\n"); + exit(1); + } + + return (ret); +} + +/* + * Display an out of memory error message and abort the current program. + */ +void +zpool_no_memory(void) +{ + assert(errno == ENOMEM); + (void) fprintf(stderr, + gettext("internal error: out of memory\n")); + exit(1); +} + +/* + * Return the number of logs in supplied nvlist + */ +uint_t +num_logs(nvlist_t *nv) +{ + uint_t nlogs = 0; + uint_t c, children; + nvlist_t **child; + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) + return (0); + + for (c = 0; c < children; c++) { + uint64_t is_log = B_FALSE; + + (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, + &is_log); + if (is_log) + nlogs++; + } + return (nlogs); +} diff --git a/cmd/zpool/zpool_util.h b/cmd/zpool/zpool_util.h new file mode 100644 index 000000000..e82f3202a --- /dev/null +++ b/cmd/zpool/zpool_util.h @@ -0,0 +1,72 @@ +/* + * 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 http://www.opensolaris.org/os/licensing. + * 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 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +#ifndef ZPOOL_UTIL_H +#define ZPOOL_UTIL_H + +#include <libnvpair.h> +#include <libzfs.h> + +#ifdef __cplusplus +extern "C" { +#endif + +/* + * Basic utility functions + */ +void *safe_malloc(size_t); +char *safe_strdup(const char *); +void zpool_no_memory(void); +uint_t num_logs(nvlist_t *nv); + +/* + * Virtual device functions + */ + +nvlist_t *make_root_vdev(zpool_handle_t *zhp, int force, int check_rep, + boolean_t isreplace, boolean_t dryrun, int argc, char **argv); + +/* + * Pool list functions + */ +int for_each_pool(int, char **, boolean_t unavail, zprop_list_t **, + zpool_iter_f, void *); + +typedef struct zpool_list zpool_list_t; + +zpool_list_t *pool_list_get(int, char **, zprop_list_t **, int *); +void pool_list_update(zpool_list_t *); +int pool_list_iter(zpool_list_t *, int unavail, zpool_iter_f, void *); +void pool_list_free(zpool_list_t *); +int pool_list_count(zpool_list_t *); +void pool_list_remove(zpool_list_t *, zpool_handle_t *); + +libzfs_handle_t *g_zfs; + +#ifdef __cplusplus +} +#endif + +#endif /* ZPOOL_UTIL_H */ diff --git a/cmd/zpool/zpool_vdev.c b/cmd/zpool/zpool_vdev.c new file mode 100644 index 000000000..10007c149 --- /dev/null +++ b/cmd/zpool/zpool_vdev.c @@ -0,0 +1,1401 @@ +/* + * 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 http://www.opensolaris.org/os/licensing. + * 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 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. + */ + +/* + * Functions to convert between a list of vdevs and an nvlist representing the + * configuration. Each entry in the list can be one of: + * + * Device vdevs + * disk=(path=..., devid=...) + * file=(path=...) + * + * Group vdevs + * raidz[1|2]=(...) + * mirror=(...) + * + * Hot spares + * + * While the underlying implementation supports it, group vdevs cannot contain + * other group vdevs. All userland verification of devices is contained within + * this file. If successful, the nvlist returned can be passed directly to the + * kernel; we've done as much verification as possible in userland. + * + * Hot spares are a special case, and passed down as an array of disk vdevs, at + * the same level as the root of the vdev tree. + * + * The only function exported by this file is 'make_root_vdev'. The + * function performs several passes: + * + * 1. Construct the vdev specification. Performs syntax validation and + * makes sure each device is valid. + * 2. Check for devices in use. Using libdiskmgt, makes sure that no + * devices are also in use. Some can be overridden using the 'force' + * flag, others cannot. + * 3. Check for replication errors if the 'force' flag is not specified. + * validates that the replication level is consistent across the + * entire pool. + * 4. Call libzfs to label any whole disks with an EFI label. + */ + +#include <assert.h> +#include <devid.h> +#include <errno.h> +#include <fcntl.h> +#include <libdiskmgt.h> +#include <libintl.h> +#include <libnvpair.h> +#include <stdio.h> +#include <string.h> +#include <unistd.h> +#include <sys/efi_partition.h> +#include <sys/stat.h> +#include <sys/vtoc.h> +#include <sys/mntent.h> + +#include "zpool_util.h" + +#define DISK_ROOT "/dev/dsk" +#define RDISK_ROOT "/dev/rdsk" +#define BACKUP_SLICE "s2" + +/* + * For any given vdev specification, we can have multiple errors. The + * vdev_error() function keeps track of whether we have seen an error yet, and + * prints out a header if its the first error we've seen. + */ +boolean_t error_seen; +boolean_t is_force; + +/*PRINTFLIKE1*/ +static void +vdev_error(const char *fmt, ...) +{ + va_list ap; + + if (!error_seen) { + (void) fprintf(stderr, gettext("invalid vdev specification\n")); + if (!is_force) + (void) fprintf(stderr, gettext("use '-f' to override " + "the following errors:\n")); + else + (void) fprintf(stderr, gettext("the following errors " + "must be manually repaired:\n")); + error_seen = B_TRUE; + } + + va_start(ap, fmt); + (void) vfprintf(stderr, fmt, ap); + va_end(ap); +} + +static void +libdiskmgt_error(int error) +{ + /* + * ENXIO/ENODEV is a valid error message if the device doesn't live in + * /dev/dsk. Don't bother printing an error message in this case. + */ + if (error == ENXIO || error == ENODEV) + return; + + (void) fprintf(stderr, gettext("warning: device in use checking " + "failed: %s\n"), strerror(error)); +} + +/* + * Validate a device, passing the bulk of the work off to libdiskmgt. + */ +static int +check_slice(const char *path, int force, boolean_t wholedisk, boolean_t isspare) +{ + char *msg; + int error = 0; + dm_who_type_t who; + + if (force) + who = DM_WHO_ZPOOL_FORCE; + else if (isspare) + who = DM_WHO_ZPOOL_SPARE; + else + who = DM_WHO_ZPOOL; + + if (dm_inuse((char *)path, &msg, who, &error) || error) { + if (error != 0) { + libdiskmgt_error(error); + return (0); + } else { + vdev_error("%s", msg); + free(msg); + return (-1); + } + } + + /* + * If we're given a whole disk, ignore overlapping slices since we're + * about to label it anyway. + */ + error = 0; + if (!wholedisk && !force && + (dm_isoverlapping((char *)path, &msg, &error) || error)) { + if (error == 0) { + /* dm_isoverlapping returned -1 */ + vdev_error(gettext("%s overlaps with %s\n"), path, msg); + free(msg); + return (-1); + } else if (error != ENODEV) { + /* libdiskmgt's devcache only handles physical drives */ + libdiskmgt_error(error); + return (0); + } + } + + return (0); +} + + +/* + * Validate a whole disk. Iterate over all slices on the disk and make sure + * that none is in use by calling check_slice(). + */ +static int +check_disk(const char *name, dm_descriptor_t disk, int force, int isspare) +{ + dm_descriptor_t *drive, *media, *slice; + int err = 0; + int i; + int ret; + + /* + * Get the drive associated with this disk. This should never fail, + * because we already have an alias handle open for the device. + */ + if ((drive = dm_get_associated_descriptors(disk, DM_DRIVE, + &err)) == NULL || *drive == NULL) { + if (err) + libdiskmgt_error(err); + return (0); + } + + if ((media = dm_get_associated_descriptors(*drive, DM_MEDIA, + &err)) == NULL) { + dm_free_descriptors(drive); + if (err) + libdiskmgt_error(err); + return (0); + } + + dm_free_descriptors(drive); + + /* + * It is possible that the user has specified a removable media drive, + * and the media is not present. + */ + if (*media == NULL) { + dm_free_descriptors(media); + vdev_error(gettext("'%s' has no media in drive\n"), name); + return (-1); + } + + if ((slice = dm_get_associated_descriptors(*media, DM_SLICE, + &err)) == NULL) { + dm_free_descriptors(media); + if (err) + libdiskmgt_error(err); + return (0); + } + + dm_free_descriptors(media); + + ret = 0; + + /* + * Iterate over all slices and report any errors. We don't care about + * overlapping slices because we are using the whole disk. + */ + for (i = 0; slice[i] != NULL; i++) { + char *name = dm_get_name(slice[i], &err); + + if (check_slice(name, force, B_TRUE, isspare) != 0) + ret = -1; + + dm_free_name(name); + } + + dm_free_descriptors(slice); + return (ret); +} + +/* + * Validate a device. + */ +static int +check_device(const char *path, boolean_t force, boolean_t isspare) +{ + dm_descriptor_t desc; + int err; + char *dev; + + /* + * For whole disks, libdiskmgt does not include the leading dev path. + */ + dev = strrchr(path, '/'); + assert(dev != NULL); + dev++; + if ((desc = dm_get_descriptor_by_name(DM_ALIAS, dev, &err)) != NULL) { + err = check_disk(path, desc, force, isspare); + dm_free_descriptor(desc); + return (err); + } + + return (check_slice(path, force, B_FALSE, isspare)); +} + +/* + * Check that a file is valid. All we can do in this case is check that it's + * not in use by another pool, and not in use by swap. + */ +static int +check_file(const char *file, boolean_t force, boolean_t isspare) +{ + char *name; + int fd; + int ret = 0; + int err; + pool_state_t state; + boolean_t inuse; + + if (dm_inuse_swap(file, &err)) { + if (err) + libdiskmgt_error(err); + else + vdev_error(gettext("%s is currently used by swap. " + "Please see swap(1M).\n"), file); + return (-1); + } + + if ((fd = open(file, O_RDONLY)) < 0) + return (0); + + if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) == 0 && inuse) { + const char *desc; + + switch (state) { + case POOL_STATE_ACTIVE: + desc = gettext("active"); + break; + + case POOL_STATE_EXPORTED: + desc = gettext("exported"); + break; + + case POOL_STATE_POTENTIALLY_ACTIVE: + desc = gettext("potentially active"); + break; + + default: + desc = gettext("unknown"); + break; + } + + /* + * Allow hot spares to be shared between pools. + */ + if (state == POOL_STATE_SPARE && isspare) + return (0); + + if (state == POOL_STATE_ACTIVE || + state == POOL_STATE_SPARE || !force) { + switch (state) { + case POOL_STATE_SPARE: + vdev_error(gettext("%s is reserved as a hot " + "spare for pool %s\n"), file, name); + break; + default: + vdev_error(gettext("%s is part of %s pool " + "'%s'\n"), file, desc, name); + break; + } + ret = -1; + } + + free(name); + } + + (void) close(fd); + return (ret); +} + + +/* + * By "whole disk" we mean an entire physical disk (something we can + * label, toggle the write cache on, etc.) as opposed to the full + * capacity of a pseudo-device such as lofi or did. We act as if we + * are labeling the disk, which should be a pretty good test of whether + * it's a viable device or not. Returns B_TRUE if it is and B_FALSE if + * it isn't. + */ +static boolean_t +is_whole_disk(const char *arg) +{ + struct dk_gpt *label; + int fd; + char path[MAXPATHLEN]; + + (void) snprintf(path, sizeof (path), "%s%s%s", + RDISK_ROOT, strrchr(arg, '/'), BACKUP_SLICE); + if ((fd = open(path, O_RDWR | O_NDELAY)) < 0) + return (B_FALSE); + if (efi_alloc_and_init(fd, EFI_NUMPAR, &label) != 0) { + (void) close(fd); + return (B_FALSE); + } + efi_free(label); + (void) close(fd); + return (B_TRUE); +} + +/* + * Create a leaf vdev. Determine if this is a file or a device. If it's a + * device, fill in the device id to make a complete nvlist. Valid forms for a + * leaf vdev are: + * + * /dev/dsk/xxx Complete disk path + * /xxx Full path to file + * xxx Shorthand for /dev/dsk/xxx + */ +static nvlist_t * +make_leaf_vdev(const char *arg, uint64_t is_log) +{ + char path[MAXPATHLEN]; + struct stat64 statbuf; + nvlist_t *vdev = NULL; + char *type = NULL; + boolean_t wholedisk = B_FALSE; + + /* + * Determine what type of vdev this is, and put the full path into + * 'path'. We detect whether this is a device of file afterwards by + * checking the st_mode of the file. + */ + if (arg[0] == '/') { + /* + * Complete device or file path. Exact type is determined by + * examining the file descriptor afterwards. + */ + wholedisk = is_whole_disk(arg); + if (!wholedisk && (stat64(arg, &statbuf) != 0)) { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + arg, strerror(errno)); + return (NULL); + } + + (void) strlcpy(path, arg, sizeof (path)); + } else { + /* + * This may be a short path for a device, or it could be total + * gibberish. Check to see if it's a known device in + * /dev/dsk/. As part of this check, see if we've been given a + * an entire disk (minus the slice number). + */ + (void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, + arg); + wholedisk = is_whole_disk(path); + if (!wholedisk && (stat64(path, &statbuf) != 0)) { + /* + * If we got ENOENT, then the user gave us + * gibberish, so try to direct them with a + * reasonable error message. Otherwise, + * regurgitate strerror() since it's the best we + * can do. + */ + if (errno == ENOENT) { + (void) fprintf(stderr, + gettext("cannot open '%s': no such " + "device in %s\n"), arg, DISK_ROOT); + (void) fprintf(stderr, + gettext("must be a full path or " + "shorthand device name\n")); + return (NULL); + } else { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + path, strerror(errno)); + return (NULL); + } + } + } + + /* + * Determine whether this is a device or a file. + */ + if (wholedisk || S_ISBLK(statbuf.st_mode)) { + type = VDEV_TYPE_DISK; + } else if (S_ISREG(statbuf.st_mode)) { + type = VDEV_TYPE_FILE; + } else { + (void) fprintf(stderr, gettext("cannot use '%s': must be a " + "block device or regular file\n"), path); + return (NULL); + } + + /* + * Finally, we have the complete device or file, and we know that it is + * acceptable to use. Construct the nvlist to describe this vdev. All + * vdevs have a 'path' element, and devices also have a 'devid' element. + */ + verify(nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) == 0); + verify(nvlist_add_string(vdev, ZPOOL_CONFIG_PATH, path) == 0); + verify(nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE, type) == 0); + verify(nvlist_add_uint64(vdev, ZPOOL_CONFIG_IS_LOG, is_log) == 0); + if (strcmp(type, VDEV_TYPE_DISK) == 0) + verify(nvlist_add_uint64(vdev, ZPOOL_CONFIG_WHOLE_DISK, + (uint64_t)wholedisk) == 0); + + /* + * For a whole disk, defer getting its devid until after labeling it. + */ + if (S_ISBLK(statbuf.st_mode) && !wholedisk) { + /* + * Get the devid for the device. + */ + int fd; + ddi_devid_t devid; + char *minor = NULL, *devid_str = NULL; + + if ((fd = open(path, O_RDONLY)) < 0) { + (void) fprintf(stderr, gettext("cannot open '%s': " + "%s\n"), path, strerror(errno)); + nvlist_free(vdev); + return (NULL); + } + + if (devid_get(fd, &devid) == 0) { + if (devid_get_minor_name(fd, &minor) == 0 && + (devid_str = devid_str_encode(devid, minor)) != + NULL) { + verify(nvlist_add_string(vdev, + ZPOOL_CONFIG_DEVID, devid_str) == 0); + } + if (devid_str != NULL) + devid_str_free(devid_str); + if (minor != NULL) + devid_str_free(minor); + devid_free(devid); + } + + (void) close(fd); + } + + return (vdev); +} + +/* + * Go through and verify the replication level of the pool is consistent. + * Performs the following checks: + * + * For the new spec, verifies that devices in mirrors and raidz are the + * same size. + * + * If the current configuration already has inconsistent replication + * levels, ignore any other potential problems in the new spec. + * + * Otherwise, make sure that the current spec (if there is one) and the new + * spec have consistent replication levels. + */ +typedef struct replication_level { + char *zprl_type; + uint64_t zprl_children; + uint64_t zprl_parity; +} replication_level_t; + +#define ZPOOL_FUZZ (16 * 1024 * 1024) + +/* + * Given a list of toplevel vdevs, return the current replication level. If + * the config is inconsistent, then NULL is returned. If 'fatal' is set, then + * an error message will be displayed for each self-inconsistent vdev. + */ +static replication_level_t * +get_replication(nvlist_t *nvroot, boolean_t fatal) +{ + nvlist_t **top; + uint_t t, toplevels; + nvlist_t **child; + uint_t c, children; + nvlist_t *nv; + char *type; + replication_level_t lastrep, rep, *ret; + boolean_t dontreport; + + ret = safe_malloc(sizeof (replication_level_t)); + + verify(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, + &top, &toplevels) == 0); + + lastrep.zprl_type = NULL; + for (t = 0; t < toplevels; t++) { + uint64_t is_log = B_FALSE; + + nv = top[t]; + + /* + * For separate logs we ignore the top level vdev replication + * constraints. + */ + (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log); + if (is_log) + continue; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, + &type) == 0); + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + /* + * This is a 'file' or 'disk' vdev. + */ + rep.zprl_type = type; + rep.zprl_children = 1; + rep.zprl_parity = 0; + } else { + uint64_t vdev_size; + + /* + * This is a mirror or RAID-Z vdev. Go through and make + * sure the contents are all the same (files vs. disks), + * keeping track of the number of elements in the + * process. + * + * We also check that the size of each vdev (if it can + * be determined) is the same. + */ + rep.zprl_type = type; + rep.zprl_children = 0; + + if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) { + verify(nvlist_lookup_uint64(nv, + ZPOOL_CONFIG_NPARITY, + &rep.zprl_parity) == 0); + assert(rep.zprl_parity != 0); + } else { + rep.zprl_parity = 0; + } + + /* + * The 'dontreport' variable indicates that we've + * already reported an error for this spec, so don't + * bother doing it again. + */ + type = NULL; + dontreport = 0; + vdev_size = -1ULL; + for (c = 0; c < children; c++) { + nvlist_t *cnv = child[c]; + char *path; + struct stat64 statbuf; + uint64_t size = -1ULL; + char *childtype; + int fd, err; + + rep.zprl_children++; + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_TYPE, &childtype) == 0); + + /* + * If this is a replacing or spare vdev, then + * get the real first child of the vdev. + */ + if (strcmp(childtype, + VDEV_TYPE_REPLACING) == 0 || + strcmp(childtype, VDEV_TYPE_SPARE) == 0) { + nvlist_t **rchild; + uint_t rchildren; + + verify(nvlist_lookup_nvlist_array(cnv, + ZPOOL_CONFIG_CHILDREN, &rchild, + &rchildren) == 0); + assert(rchildren == 2); + cnv = rchild[0]; + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_TYPE, + &childtype) == 0); + } + + verify(nvlist_lookup_string(cnv, + ZPOOL_CONFIG_PATH, &path) == 0); + + /* + * If we have a raidz/mirror that combines disks + * with files, report it as an error. + */ + if (!dontreport && type != NULL && + strcmp(type, childtype) != 0) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication " + "level: %s contains both " + "files and devices\n"), + rep.zprl_type); + else + return (NULL); + dontreport = B_TRUE; + } + + /* + * According to stat(2), the value of 'st_size' + * is undefined for block devices and character + * devices. But there is no effective way to + * determine the real size in userland. + * + * Instead, we'll take advantage of an + * implementation detail of spec_size(). If the + * device is currently open, then we (should) + * return a valid size. + * + * If we still don't get a valid size (indicated + * by a size of 0 or MAXOFFSET_T), then ignore + * this device altogether. + */ + if ((fd = open(path, O_RDONLY)) >= 0) { + err = fstat64(fd, &statbuf); + (void) close(fd); + } else { + err = stat64(path, &statbuf); + } + + if (err != 0 || + statbuf.st_size == 0 || + statbuf.st_size == MAXOFFSET_T) + continue; + + size = statbuf.st_size; + + /* + * Also make sure that devices and + * slices have a consistent size. If + * they differ by a significant amount + * (~16MB) then report an error. + */ + if (!dontreport && + (vdev_size != -1ULL && + (labs(size - vdev_size) > + ZPOOL_FUZZ))) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "%s contains devices of " + "different sizes\n"), + rep.zprl_type); + else + return (NULL); + dontreport = B_TRUE; + } + + type = childtype; + vdev_size = size; + } + } + + /* + * At this point, we have the replication of the last toplevel + * vdev in 'rep'. Compare it to 'lastrep' to see if its + * different. + */ + if (lastrep.zprl_type != NULL) { + if (strcmp(lastrep.zprl_type, rep.zprl_type) != 0) { + if (ret != NULL) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %s and %s vdevs are " + "present\n"), + lastrep.zprl_type, rep.zprl_type); + else + return (NULL); + } else if (lastrep.zprl_parity != rep.zprl_parity) { + if (ret) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %llu and %llu device parity " + "%s vdevs are present\n"), + lastrep.zprl_parity, + rep.zprl_parity, + rep.zprl_type); + else + return (NULL); + } else if (lastrep.zprl_children != rep.zprl_children) { + if (ret) + free(ret); + ret = NULL; + if (fatal) + vdev_error(gettext( + "mismatched replication level: " + "both %llu-way and %llu-way %s " + "vdevs are present\n"), + lastrep.zprl_children, + rep.zprl_children, + rep.zprl_type); + else + return (NULL); + } + } + lastrep = rep; + } + + if (ret != NULL) + *ret = rep; + + return (ret); +} + +/* + * Check the replication level of the vdev spec against the current pool. Calls + * get_replication() to make sure the new spec is self-consistent. If the pool + * has a consistent replication level, then we ignore any errors. Otherwise, + * report any difference between the two. + */ +static int +check_replication(nvlist_t *config, nvlist_t *newroot) +{ + nvlist_t **child; + uint_t children; + replication_level_t *current = NULL, *new; + int ret; + + /* + * If we have a current pool configuration, check to see if it's + * self-consistent. If not, simply return success. + */ + if (config != NULL) { + nvlist_t *nvroot; + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if ((current = get_replication(nvroot, B_FALSE)) == NULL) + return (0); + } + /* + * for spares there may be no children, and therefore no + * replication level to check + */ + if ((nvlist_lookup_nvlist_array(newroot, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) || (children == 0)) { + free(current); + return (0); + } + + /* + * If all we have is logs then there's no replication level to check. + */ + if (num_logs(newroot) == children) { + free(current); + return (0); + } + + /* + * Get the replication level of the new vdev spec, reporting any + * inconsistencies found. + */ + if ((new = get_replication(newroot, B_TRUE)) == NULL) { + free(current); + return (-1); + } + + /* + * Check to see if the new vdev spec matches the replication level of + * the current pool. + */ + ret = 0; + if (current != NULL) { + if (strcmp(current->zprl_type, new->zprl_type) != 0) { + vdev_error(gettext( + "mismatched replication level: pool uses %s " + "and new vdev is %s\n"), + current->zprl_type, new->zprl_type); + ret = -1; + } else if (current->zprl_parity != new->zprl_parity) { + vdev_error(gettext( + "mismatched replication level: pool uses %llu " + "device parity and new vdev uses %llu\n"), + current->zprl_parity, new->zprl_parity); + ret = -1; + } else if (current->zprl_children != new->zprl_children) { + vdev_error(gettext( + "mismatched replication level: pool uses %llu-way " + "%s and new vdev uses %llu-way %s\n"), + current->zprl_children, current->zprl_type, + new->zprl_children, new->zprl_type); + ret = -1; + } + } + + free(new); + if (current != NULL) + free(current); + + return (ret); +} + +/* + * Go through and find any whole disks in the vdev specification, labelling them + * as appropriate. When constructing the vdev spec, we were unable to open this + * device in order to provide a devid. Now that we have labelled the disk and + * know that slice 0 is valid, we can construct the devid now. + * + * If the disk was already labeled with an EFI label, we will have gotten the + * devid already (because we were able to open the whole disk). Otherwise, we + * need to get the devid after we label the disk. + */ +static int +make_disks(zpool_handle_t *zhp, nvlist_t *nv) +{ + nvlist_t **child; + uint_t c, children; + char *type, *path, *diskname; + char buf[MAXPATHLEN]; + uint64_t wholedisk; + int fd; + int ret; + ddi_devid_t devid; + char *minor = NULL, *devid_str = NULL; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + + if (strcmp(type, VDEV_TYPE_DISK) != 0) + return (0); + + /* + * We have a disk device. Get the path to the device + * and see if it's a whole disk by appending the backup + * slice and stat()ing the device. + */ + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, + &wholedisk) != 0 || !wholedisk) + return (0); + + diskname = strrchr(path, '/'); + assert(diskname != NULL); + diskname++; + if (zpool_label_disk(g_zfs, zhp, diskname) == -1) + return (-1); + + /* + * Fill in the devid, now that we've labeled the disk. + */ + (void) snprintf(buf, sizeof (buf), "%ss0", path); + if ((fd = open(buf, O_RDONLY)) < 0) { + (void) fprintf(stderr, + gettext("cannot open '%s': %s\n"), + buf, strerror(errno)); + return (-1); + } + + if (devid_get(fd, &devid) == 0) { + if (devid_get_minor_name(fd, &minor) == 0 && + (devid_str = devid_str_encode(devid, minor)) != + NULL) { + verify(nvlist_add_string(nv, + ZPOOL_CONFIG_DEVID, devid_str) == 0); + } + if (devid_str != NULL) + devid_str_free(devid_str); + if (minor != NULL) + devid_str_free(minor); + devid_free(devid); + } + + /* + * Update the path to refer to the 's0' slice. The presence of + * the 'whole_disk' field indicates to the CLI that we should + * chop off the slice number when displaying the device in + * future output. + */ + verify(nvlist_add_string(nv, ZPOOL_CONFIG_PATH, buf) == 0); + + (void) close(fd); + + return (0); + } + + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = make_disks(zhp, child[c])) != 0) + return (ret); + + return (0); +} + +/* + * Determine if the given path is a hot spare within the given configuration. + */ +static boolean_t +is_spare(nvlist_t *config, const char *path) +{ + int fd; + pool_state_t state; + char *name = NULL; + nvlist_t *label; + uint64_t guid, spareguid; + nvlist_t *nvroot; + nvlist_t **spares; + uint_t i, nspares; + boolean_t inuse; + + if ((fd = open(path, O_RDONLY)) < 0) + return (B_FALSE); + + if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) != 0 || + !inuse || + state != POOL_STATE_SPARE || + zpool_read_label(fd, &label) != 0) { + free(name); + (void) close(fd); + return (B_FALSE); + } + free(name); + + (void) close(fd); + verify(nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) == 0); + nvlist_free(label); + + verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, + &nvroot) == 0); + if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + &spares, &nspares) == 0) { + for (i = 0; i < nspares; i++) { + verify(nvlist_lookup_uint64(spares[i], + ZPOOL_CONFIG_GUID, &spareguid) == 0); + if (spareguid == guid) + return (B_TRUE); + } + } + + return (B_FALSE); +} + +/* + * Go through and find any devices that are in use. We rely on libdiskmgt for + * the majority of this task. + */ +static int +check_in_use(nvlist_t *config, nvlist_t *nv, int force, int isreplacing, + int isspare) +{ + nvlist_t **child; + uint_t c, children; + char *type, *path; + int ret; + char buf[MAXPATHLEN]; + uint64_t wholedisk; + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, + &child, &children) != 0) { + + verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); + + /* + * As a generic check, we look to see if this is a replace of a + * hot spare within the same pool. If so, we allow it + * regardless of what libdiskmgt or zpool_in_use() says. + */ + if (isreplacing) { + if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, + &wholedisk) == 0 && wholedisk) + (void) snprintf(buf, sizeof (buf), "%ss0", + path); + else + (void) strlcpy(buf, path, sizeof (buf)); + if (is_spare(config, buf)) + return (0); + } + + if (strcmp(type, VDEV_TYPE_DISK) == 0) + ret = check_device(path, force, isspare); + + if (strcmp(type, VDEV_TYPE_FILE) == 0) + ret = check_file(path, force, isspare); + + return (ret); + } + + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_FALSE)) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_TRUE)) != 0) + return (ret); + + if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, + &child, &children) == 0) + for (c = 0; c < children; c++) + if ((ret = check_in_use(config, child[c], force, + isreplacing, B_FALSE)) != 0) + return (ret); + + return (0); +} + +static const char * +is_grouping(const char *type, int *mindev) +{ + if (strcmp(type, "raidz") == 0 || strcmp(type, "raidz1") == 0) { + if (mindev != NULL) + *mindev = 2; + return (VDEV_TYPE_RAIDZ); + } + + if (strcmp(type, "raidz2") == 0) { + if (mindev != NULL) + *mindev = 3; + return (VDEV_TYPE_RAIDZ); + } + + if (strcmp(type, "mirror") == 0) { + if (mindev != NULL) + *mindev = 2; + return (VDEV_TYPE_MIRROR); + } + + if (strcmp(type, "spare") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_SPARE); + } + + if (strcmp(type, "log") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_LOG); + } + + if (strcmp(type, "cache") == 0) { + if (mindev != NULL) + *mindev = 1; + return (VDEV_TYPE_L2CACHE); + } + + return (NULL); +} + +/* + * Construct a syntactically valid vdev specification, + * and ensure that all devices and files exist and can be opened. + * Note: we don't bother freeing anything in the error paths + * because the program is just going to exit anyway. + */ +nvlist_t * +construct_spec(int argc, char **argv) +{ + nvlist_t *nvroot, *nv, **top, **spares, **l2cache; + int t, toplevels, mindev, nspares, nlogs, nl2cache; + const char *type; + uint64_t is_log; + boolean_t seen_logs; + + top = NULL; + toplevels = 0; + spares = NULL; + l2cache = NULL; + nspares = 0; + nlogs = 0; + nl2cache = 0; + is_log = B_FALSE; + seen_logs = B_FALSE; + + while (argc > 0) { + nv = NULL; + + /* + * If it's a mirror or raidz, the subsequent arguments are + * its leaves -- until we encounter the next mirror or raidz. + */ + if ((type = is_grouping(argv[0], &mindev)) != NULL) { + nvlist_t **child = NULL; + int c, children = 0; + + if (strcmp(type, VDEV_TYPE_SPARE) == 0) { + if (spares != NULL) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'spare' can be " + "specified only once\n")); + return (NULL); + } + is_log = B_FALSE; + } + + if (strcmp(type, VDEV_TYPE_LOG) == 0) { + if (seen_logs) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'log' can be " + "specified only once\n")); + return (NULL); + } + seen_logs = B_TRUE; + is_log = B_TRUE; + argc--; + argv++; + /* + * A log is not a real grouping device. + * We just set is_log and continue. + */ + continue; + } + + if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) { + if (l2cache != NULL) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: 'cache' can be " + "specified only once\n")); + return (NULL); + } + is_log = B_FALSE; + } + + if (is_log) { + if (strcmp(type, VDEV_TYPE_MIRROR) != 0) { + (void) fprintf(stderr, + gettext("invalid vdev " + "specification: unsupported 'log' " + "device: %s\n"), type); + return (NULL); + } + nlogs++; + } + + for (c = 1; c < argc; c++) { + if (is_grouping(argv[c], NULL) != NULL) + break; + children++; + child = realloc(child, + children * sizeof (nvlist_t *)); + if (child == NULL) + zpool_no_memory(); + if ((nv = make_leaf_vdev(argv[c], B_FALSE)) + == NULL) + return (NULL); + child[children - 1] = nv; + } + + if (children < mindev) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: %s requires at least %d " + "devices\n"), argv[0], mindev); + return (NULL); + } + + argc -= c; + argv += c; + + if (strcmp(type, VDEV_TYPE_SPARE) == 0) { + spares = child; + nspares = children; + continue; + } else if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) { + l2cache = child; + nl2cache = children; + continue; + } else { + verify(nvlist_alloc(&nv, NV_UNIQUE_NAME, + 0) == 0); + verify(nvlist_add_string(nv, ZPOOL_CONFIG_TYPE, + type) == 0); + verify(nvlist_add_uint64(nv, + ZPOOL_CONFIG_IS_LOG, is_log) == 0); + if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) { + verify(nvlist_add_uint64(nv, + ZPOOL_CONFIG_NPARITY, + mindev - 1) == 0); + } + verify(nvlist_add_nvlist_array(nv, + ZPOOL_CONFIG_CHILDREN, child, + children) == 0); + + for (c = 0; c < children; c++) + nvlist_free(child[c]); + free(child); + } + } else { + /* + * We have a device. Pass off to make_leaf_vdev() to + * construct the appropriate nvlist describing the vdev. + */ + if ((nv = make_leaf_vdev(argv[0], is_log)) == NULL) + return (NULL); + if (is_log) + nlogs++; + argc--; + argv++; + } + + toplevels++; + top = realloc(top, toplevels * sizeof (nvlist_t *)); + if (top == NULL) + zpool_no_memory(); + top[toplevels - 1] = nv; + } + + if (toplevels == 0 && nspares == 0 && nl2cache == 0) { + (void) fprintf(stderr, gettext("invalid vdev " + "specification: at least one toplevel vdev must be " + "specified\n")); + return (NULL); + } + + if (seen_logs && nlogs == 0) { + (void) fprintf(stderr, gettext("invalid vdev specification: " + "log requires at least 1 device\n")); + return (NULL); + } + + /* + * Finally, create nvroot and add all top-level vdevs to it. + */ + verify(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) == 0); + verify(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, + VDEV_TYPE_ROOT) == 0); + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, + top, toplevels) == 0); + if (nspares != 0) + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, + spares, nspares) == 0); + if (nl2cache != 0) + verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, + l2cache, nl2cache) == 0); + + for (t = 0; t < toplevels; t++) + nvlist_free(top[t]); + for (t = 0; t < nspares; t++) + nvlist_free(spares[t]); + for (t = 0; t < nl2cache; t++) + nvlist_free(l2cache[t]); + if (spares) + free(spares); + if (l2cache) + free(l2cache); + free(top); + + return (nvroot); +} + + +/* + * Get and validate the contents of the given vdev specification. This ensures + * that the nvlist returned is well-formed, that all the devices exist, and that + * they are not currently in use by any other known consumer. The 'poolconfig' + * parameter is the current configuration of the pool when adding devices + * existing pool, and is used to perform additional checks, such as changing the + * replication level of the pool. It can be 'NULL' to indicate that this is a + * new pool. The 'force' flag controls whether devices should be forcefully + * added, even if they appear in use. + */ +nvlist_t * +make_root_vdev(zpool_handle_t *zhp, int force, int check_rep, + boolean_t isreplacing, boolean_t dryrun, int argc, char **argv) +{ + nvlist_t *newroot; + nvlist_t *poolconfig = NULL; + is_force = force; + + /* + * Construct the vdev specification. If this is successful, we know + * that we have a valid specification, and that all devices can be + * opened. + */ + if ((newroot = construct_spec(argc, argv)) == NULL) + return (NULL); + + if (zhp && ((poolconfig = zpool_get_config(zhp, NULL)) == NULL)) + return (NULL); + + /* + * Validate each device to make sure that its not shared with another + * subsystem. We do this even if 'force' is set, because there are some + * uses (such as a dedicated dump device) that even '-f' cannot + * override. + */ + if (check_in_use(poolconfig, newroot, force, isreplacing, + B_FALSE) != 0) { + nvlist_free(newroot); + return (NULL); + } + + /* + * Check the replication level of the given vdevs and report any errors + * found. We include the existing pool spec, if any, as we need to + * catch changes against the existing replication level. + */ + if (check_rep && check_replication(poolconfig, newroot) != 0) { + nvlist_free(newroot); + return (NULL); + } + + /* + * Run through the vdev specification and label any whole disks found. + */ + if (!dryrun && make_disks(zhp, newroot) != 0) { + nvlist_free(newroot); + return (NULL); + } + + return (newroot); +} |