Rebase master to b121

cmd/zdb/zdb.c

@@ -818,6 +818,8 @@ dump_dsl_dataset(objset_t *os, uint64_t object, void *data, size_t size)
 	    (u_longlong_t)ds->ds_snapnames_zapobj);
 	(void) printf("\t\tnum_children = %llu\n",
 	    (u_longlong_t)ds->ds_num_children);
+	(void) printf("\t\tuserrefs_obj = %llu\n",
+	    (u_longlong_t)ds->ds_userrefs_obj);
 	(void) printf("\t\tcreation_time = %s", ctime(&crtime));
 	(void) printf("\t\tcreation_txg = %llu\n",
 	    (u_longlong_t)ds->ds_creation_txg);
@@ -1049,6 +1051,7 @@ static object_viewer_t *object_viewer[DMU_OT_NUMTYPES] = {
 	dump_zap,		/* DSL scrub queue		*/
 	dump_zap,		/* ZFS user/group used		*/
 	dump_zap,		/* ZFS user/group quota		*/
+	dump_zap,		/* snapshot refcount tags	*/
 };
 
 static void

cmd/zfs/zfs_iter.c

@@ -362,7 +362,7 @@ zfs_for_each(int argc, char **argv, int flags, zfs_type_t types,
 	cb.cb_types = types;
 	cb.cb_depth_limit = limit;
 	/*
-	 * If cb_proplist is provided then in the zfs_handles created  we
+	 * If cb_proplist is provided then in the zfs_handles created we
 	 * retain only those properties listed in cb_proplist and sortcol.
 	 * The rest are pruned. So, the caller should make sure that no other
 	 * properties other than those listed in cb_proplist/sortcol are

cmd/zfs/zfs_main.c

@@ -80,6 +80,8 @@ static int zfs_do_receive(int argc, char **argv);
 static int zfs_do_promote(int argc, char **argv);
 static int zfs_do_userspace(int argc, char **argv);
 static int zfs_do_python(int argc, char **argv);
+static int zfs_do_hold(int argc, char **argv);
+static int zfs_do_release(int argc, char **argv);
 
 /*
  * Enable a reasonable set of defaults for libumem debugging on DEBUG builds.
@@ -121,7 +123,10 @@ typedef enum {
 	HELP_ALLOW,
 	HELP_UNALLOW,
 	HELP_USERSPACE,
-	HELP_GROUPSPACE
+	HELP_GROUPSPACE,
+	HELP_HOLD,
+	HELP_HOLDS,
+	HELP_RELEASE
 } zfs_help_t;
 
 typedef struct zfs_command {
@@ -169,6 +174,10 @@ static zfs_command_t command_table[] = {
 	{ "allow",	zfs_do_python,		HELP_ALLOW		},
 	{ NULL },
 	{ "unallow",	zfs_do_python,		HELP_UNALLOW		},
+	{ NULL },
+	{ "hold",	zfs_do_hold,		HELP_HOLD		},
+	{ "holds",	zfs_do_python,		HELP_HOLDS		},
+	{ "release",	zfs_do_release,		HELP_RELEASE		},
 };
 
 #define	NCOMMAND	(sizeof (command_table) / sizeof (command_table[0]))
@@ -189,7 +198,8 @@ get_usage(zfs_help_t idx)
 		    "-V <size> <volume>\n"));
 	case HELP_DESTROY:
 		return (gettext("\tdestroy [-rRf] "
-		    "<filesystem|volume|snapshot>\n"));
+		    "<filesystem|volume|snapshot>\n"
+		    "\tdestroy -d [-r] <filesystem|volume|snapshot>\n"));
 	case HELP_GET:
 		return (gettext("\tget [-rHp] [-d max] "
 		    "[-o field[,...]] [-s source[,...]]\n"
@@ -236,7 +246,7 @@ get_usage(zfs_help_t idx)
 		return (gettext("\tunmount [-f] "
 		    "<-a | filesystem|mountpoint>\n"));
 	case HELP_UNSHARE:
-		return (gettext("\tunshare [-f] "
+		return (gettext("\tunshare "
 		    "<-a | filesystem|mountpoint>\n"));
 	case HELP_ALLOW:
 		return (gettext("\tallow <filesystem|volume>\n"
@@ -266,6 +276,12 @@ get_usage(zfs_help_t idx)
 		return (gettext("\tgroupspace [-hniHpU] [-o field[,...]] "
 		    "[-sS field] ... [-t type[,...]]\n"
 		    "\t    <filesystem|snapshot>\n"));
+	case HELP_HOLD:
+		return (gettext("\thold [-r] <tag> <snapshot> ...\n"));
+	case HELP_HOLDS:
+		return (gettext("\tholds [-r] <snapshot> ...\n"));
+	case HELP_RELEASE:
+		return (gettext("\trelease [-r] <tag> <snapshot> ...\n"));
 	}
 
 	abort();
@@ -769,11 +785,13 @@ badusage:
 }
 
 /*
- * zfs destroy [-rf] <fs, snap, vol>
+ * zfs destroy [-rRf] <fs, snap, vol>
+ * zfs destroy -d [-r] <fs, snap, vol>
  *
  * 	-r	Recursively destroy all children
  * 	-R	Recursively destroy all dependents, including clones
  * 	-f	Force unmounting of any dependents
+ *	-d	If we can't destroy now, mark for deferred destruction
  *
  * Destroys the given dataset.  By default, it will unmount any filesystems,
  * and refuse to destroy a dataset that has any dependents.  A dependent can
@@ -789,6 +807,7 @@ typedef struct destroy_cbdata {
 	boolean_t	cb_closezhp;
 	zfs_handle_t	*cb_target;
 	char		*cb_snapname;
+	boolean_t	cb_defer_destroy;
 } destroy_cbdata_t;
 
 /*
@@ -869,7 +888,7 @@ destroy_callback(zfs_handle_t *zhp, void *data)
 	 * Bail out on the first error.
 	 */
 	if (zfs_unmount(zhp, NULL, cbp->cb_force ? MS_FORCE : 0) != 0 ||
-	    zfs_destroy(zhp) != 0) {
+	    zfs_destroy(zhp, cbp->cb_defer_destroy) != 0) {
 		zfs_close(zhp);
 		return (-1);
 	}
@@ -923,8 +942,11 @@ zfs_do_destroy(int argc, char **argv)
 	char *cp;
 
 	/* check options */
-	while ((c = getopt(argc, argv, "frR")) != -1) {
+	while ((c = getopt(argc, argv, "dfrR")) != -1) {
 		switch (c) {
+		case 'd':
+			cb.cb_defer_destroy = B_TRUE;
+			break;
 		case 'f':
 			cb.cb_force = 1;
 			break;
@@ -956,6 +978,9 @@ zfs_do_destroy(int argc, char **argv)
 		usage(B_FALSE);
 	}
 
+	if (cb.cb_defer_destroy && cb.cb_doclones)
+		usage(B_FALSE);
+
 	/*
 	 * If we are doing recursive destroy of a snapshot, then the
 	 * named snapshot may not exist.  Go straight to libzfs.
@@ -977,7 +1002,7 @@ zfs_do_destroy(int argc, char **argv)
 			}
 		}
 
-		ret = zfs_destroy_snaps(zhp, cp);
+		ret = zfs_destroy_snaps(zhp, cp, cb.cb_defer_destroy);
 		zfs_close(zhp);
 		if (ret) {
 			(void) fprintf(stderr,
@@ -986,7 +1011,6 @@ zfs_do_destroy(int argc, char **argv)
 		return (ret != 0);
 	}
 
-
 	/* Open the given dataset */
 	if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET)) == NULL)
 		return (1);
@@ -1014,15 +1038,15 @@ zfs_do_destroy(int argc, char **argv)
 	 * Check for any dependents and/or clones.
 	 */
 	cb.cb_first = B_TRUE;
-	if (!cb.cb_doclones &&
+	if (!cb.cb_doclones && !cb.cb_defer_destroy &&
 	    zfs_iter_dependents(zhp, B_TRUE, destroy_check_dependent,
 	    &cb) != 0) {
 		zfs_close(zhp);
 		return (1);
 	}
 
-	if (cb.cb_error ||
-	    zfs_iter_dependents(zhp, B_FALSE, destroy_callback, &cb) != 0) {
+	if (cb.cb_error || (!cb.cb_defer_destroy &&
+	    (zfs_iter_dependents(zhp, B_FALSE, destroy_callback, &cb) != 0))) {
 		zfs_close(zhp);
 		return (1);
 	}
@@ -1035,7 +1059,6 @@ zfs_do_destroy(int argc, char **argv)
 	if (destroy_callback(zhp, &cb) != 0)
 		return (1);
 
-
 	return (0);
 }
 
@@ -1613,7 +1636,7 @@ zfs_do_upgrade(int argc, char **argv)
 		(void) printf(gettext(" 1   Initial ZFS filesystem version\n"));
 		(void) printf(gettext(" 2   Enhanced directory entries\n"));
 		(void) printf(gettext(" 3   Case insensitive and File system "
-		    "unique identifer (FUID)\n"));
+		    "unique identifier (FUID)\n"));
 		(void) printf(gettext(" 4   userquota, groupquota "
 		    "properties\n"));
 		(void) printf(gettext("\nFor more information on a particular "
@@ -2651,6 +2674,108 @@ zfs_do_receive(int argc, char **argv)
 	return (err != 0);
 }
 
+static int
+zfs_do_hold_rele_impl(int argc, char **argv, boolean_t holding)
+{
+	int errors = 0;
+	int i;
+	const char *tag;
+	boolean_t recursive = B_FALSE;
+	int c;
+	int (*func)(zfs_handle_t *, const char *, const char *, boolean_t);
+
+	/* check options */
+	while ((c = getopt(argc, argv, "r")) != -1) {
+		switch (c) {
+		case 'r':
+			recursive = B_TRUE;
+			break;
+		case '?':
+			(void) fprintf(stderr, gettext("invalid option '%c'\n"),
+			    optopt);
+			usage(B_FALSE);
+		}
+	}
+
+	argc -= optind;
+	argv += optind;
+
+	/* check number of arguments */
+	if (argc < 2)
+		usage(B_FALSE);
+
+	tag = argv[0];
+	--argc;
+	++argv;
+
+	if (holding) {
+		if (tag[0] == '.') {
+			/* tags starting with '.' are reserved for libzfs */
+			(void) fprintf(stderr,
+			    gettext("tag may not start with '.'\n"));
+			usage(B_FALSE);
+		}
+		func = zfs_hold;
+	} else {
+		func = zfs_release;
+	}
+
+	for (i = 0; i < argc; ++i) {
+		zfs_handle_t *zhp;
+		char parent[ZFS_MAXNAMELEN];
+		const char *delim;
+		char *path = argv[i];
+
+		delim = strchr(path, '@');
+		if (delim == NULL) {
+			(void) fprintf(stderr,
+			    gettext("'%s' is not a snapshot\n"), path);
+			++errors;
+			continue;
+		}
+		(void) strncpy(parent, path, delim - path);
+		parent[delim - path] = '\0';
+
+		zhp = zfs_open(g_zfs, parent,
+		    ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME);
+		if (zhp == NULL) {
+			++errors;
+			continue;
+		}
+		if (func(zhp, delim+1, tag, recursive) != 0)
+			++errors;
+		zfs_close(zhp);
+	}
+
+	return (errors != 0);
+}
+
+/*
+ * zfs hold [-r] <tag> <snap> ...
+ *
+ * 	-r	Recursively hold
+ *
+ * Apply a user-hold with the given tag to the list of snapshots.
+ */
+static int
+zfs_do_hold(int argc, char **argv)
+{
+	return (zfs_do_hold_rele_impl(argc, argv, B_TRUE));
+}
+
+/*
+ * zfs release [-r] <tag> <snap> ...
+ *
+ * 	-r	Recursively release
+ *
+ * Release a user-hold with the given tag from the list of snapshots.
+ */
+static int
+zfs_do_release(int argc, char **argv)
+{
+	return (zfs_do_hold_rele_impl(argc, argv, B_FALSE));
+}
+
 typedef struct get_all_cbdata {
 	zfs_handle_t	**cb_handles;
 	size_t		cb_alloc;

cmd/zpool/zpool_main.c

@@ -1469,7 +1469,7 @@ show_import(nvlist_t *config)
  */
 static int
 do_import(nvlist_t *config, const char *newname, const char *mntopts,
-    int force, nvlist_t *props, boolean_t allowfaulted)
+    int force, nvlist_t *props, boolean_t do_verbatim)
 {
 	zpool_handle_t *zhp;
 	char *name;
@@ -1522,14 +1522,14 @@ do_import(nvlist_t *config, const char *newname, const char *mntopts,
 		}
 	}
 
-	if (zpool_import_props(g_zfs, config, newname, props,
-	    allowfaulted) != 0)
+	if (zpool_import_props(g_zfs, config, newname, props, do_verbatim) != 0)
 		return (1);
 
 	if (newname != NULL)
 		name = (char *)newname;
 
-	verify((zhp = zpool_open_canfail(g_zfs, name)) != NULL);
+	if ((zhp = zpool_open_canfail(g_zfs, name)) == NULL)
+		return (1);
 
 	if (zpool_get_state(zhp) != POOL_STATE_UNAVAIL &&
 	    zpool_enable_datasets(zhp, mntopts, 0) != 0) {
@@ -1566,7 +1566,8 @@ do_import(nvlist_t *config, const char *newname, const char *mntopts,
  *       -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.
+ *		vdevs in the FAULTED state. In other words, it does verbatim
+ *		import.
  *
  *       -a	Import all pools found.
  *
@@ -1595,7 +1596,7 @@ zpool_do_import(int argc, char **argv)
 	nvlist_t *found_config;
 	nvlist_t *props = NULL;
 	boolean_t first;
-	boolean_t allow_faulted = B_FALSE;
+	boolean_t do_verbatim = B_FALSE;
 	uint64_t pool_state;
 	char *cachefile = NULL;
 
@@ -1628,7 +1629,7 @@ zpool_do_import(int argc, char **argv)
 			do_force = B_TRUE;
 			break;
 		case 'F':
-			allow_faulted = B_TRUE;
+			do_verbatim = B_TRUE;
 			break;
 		case 'o':
 			if ((propval = strchr(optarg, '=')) != NULL) {
@@ -1778,7 +1779,7 @@ zpool_do_import(int argc, char **argv)
 
 			if (do_all)
 				err |= do_import(config, NULL, mntopts,
-				    do_force, props, allow_faulted);
+				    do_force, props, do_verbatim);
 			else
 				show_import(config);
 		} else if (searchname != NULL) {
@@ -1826,7 +1827,7 @@ zpool_do_import(int argc, char **argv)
 			err = B_TRUE;
 		} else {
 			err |= do_import(found_config, argc == 1 ? NULL :
-			    argv[1], mntopts, do_force, props, allow_faulted);
+			    argv[1], mntopts, do_force, props, do_verbatim);
 		}
 	}
 
@@ -3117,6 +3118,17 @@ status_callback(zpool_handle_t *zhp, void *data)
 		    "replace'.\n"));
 		break;
 
+	case ZPOOL_STATUS_REMOVED_DEV:
+		(void) printf(gettext("status: One or more devices has "
+		    "been removed 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 "
@@ -3539,6 +3551,8 @@ zpool_do_upgrade(int argc, char **argv)
 		(void) printf(gettext(" 14  passthrough-x aclinherit\n"));
 		(void) printf(gettext(" 15  user/group space accounting\n"));
 		(void) printf(gettext(" 16  stmf property support\n"));
+		(void) printf(gettext(" 17  Triple-parity RAID-Z\n"));
+		(void) printf(gettext(" 18  snapshot user holds\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/"
@@ -3624,6 +3638,8 @@ char *hist_event_table[LOG_END] = {
 	"refquota set",
 	"refreservation set",
 	"pool scrub done",
+	"user hold",
+	"user release",
 };
 
 /*

cmd/zpool/zpool_vdev.c

@@ -20,7 +20,7 @@
  */
 
 /*
- * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
+ * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
  */
 
@@ -67,6 +67,7 @@
 #include <libdiskmgt.h>
 #include <libintl.h>
 #include <libnvpair.h>
+#include <limits.h>
 #include <stdio.h>
 #include <string.h>
 #include <unistd.h>
@@ -1093,19 +1094,34 @@ check_in_use(nvlist_t *config, nvlist_t *nv, int force, int isreplacing,
 }
 
 static const char *
-is_grouping(const char *type, int *mindev)
+is_grouping(const char *type, int *mindev, int *maxdev)
 {
-	if (strcmp(type, "raidz") == 0 || strcmp(type, "raidz1") == 0) {
+	if (strncmp(type, "raidz", 5) == 0) {
+		const char *p = type + 5;
+		char *end;
+		long nparity;
+
+		if (*p == '\0') {
+			nparity = 1;
+		} else if (*p == '0') {
+			return (NULL); /* no zero prefixes allowed */
+		} else {
+			errno = 0;
+			nparity = strtol(p, &end, 10);
+			if (errno != 0 || nparity < 1 || nparity >= 255 ||
+			    *end != '\0')
+				return (NULL);
+		}
+
 		if (mindev != NULL)
-			*mindev = 2;
+			*mindev = nparity + 1;
+		if (maxdev != NULL)
+			*maxdev = 255;
 		return (VDEV_TYPE_RAIDZ);
 	}
 
-	if (strcmp(type, "raidz2") == 0) {
-		if (mindev != NULL)
-			*mindev = 3;
-		return (VDEV_TYPE_RAIDZ);
-	}
+	if (maxdev != NULL)
+		*maxdev = INT_MAX;
 
 	if (strcmp(type, "mirror") == 0) {
 		if (mindev != NULL)
@@ -1144,7 +1160,7 @@ nvlist_t *
 construct_spec(int argc, char **argv)
 {
 	nvlist_t *nvroot, *nv, **top, **spares, **l2cache;
-	int t, toplevels, mindev, nspares, nlogs, nl2cache;
+	int t, toplevels, mindev, maxdev, nspares, nlogs, nl2cache;
 	const char *type;
 	uint64_t is_log;
 	boolean_t seen_logs;
@@ -1166,7 +1182,7 @@ construct_spec(int argc, char **argv)
 		 * 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) {
+		if ((type = is_grouping(argv[0], &mindev, &maxdev)) != NULL) {
 			nvlist_t **child = NULL;
 			int c, children = 0;
 
@@ -1223,7 +1239,7 @@ construct_spec(int argc, char **argv)
 			}
 
 			for (c = 1; c < argc; c++) {
-				if (is_grouping(argv[c], NULL) != NULL)
+				if (is_grouping(argv[c], NULL, NULL) != NULL)
 					break;
 				children++;
 				child = realloc(child,
@@ -1243,6 +1259,13 @@ construct_spec(int argc, char **argv)
 				return (NULL);
 			}
 
+			if (children > maxdev) {
+				(void) fprintf(stderr, gettext("invalid vdev "
+				    "specification: %s supports no more than "
+				    "%d devices\n"), argv[0], maxdev);
+				return (NULL);
+			}
+
 			argc -= c;
 			argv += c;
 

cmd/ztest/ztest.c

@@ -479,7 +479,7 @@ process_options(int argc, char **argv)
 			zopt_raidz = MAX(1, value);
 			break;
 		case 'R':
-			zopt_raidz_parity = MIN(MAX(value, 1), 2);
+			zopt_raidz_parity = MIN(MAX(value, 1), 3);
 			break;
 		case 'd':
 			zopt_datasets = MAX(1, value);
@@ -1387,7 +1387,7 @@ ztest_destroy_cb(char *name, void *arg)
 	/*
 	 * Destroy the dataset.
 	 */
-	error = dmu_objset_destroy(name);
+	error = dmu_objset_destroy(name, B_FALSE);
 	if (error) {
 		(void) dmu_objset_open(name, DMU_OST_OTHER,
 		    DS_MODE_USER | DS_MODE_READONLY, &os);
@@ -1560,7 +1560,7 @@ ztest_dmu_objset_create_destroy(ztest_args_t *za)
 	zil_close(zilog);
 	dmu_objset_close(os);
 
-	error = dmu_objset_destroy(name);
+	error = dmu_objset_destroy(name, B_FALSE);
 	if (error)
 		fatal(0, "dmu_objset_destroy(%s) = %d", name, error);
 
@@ -1583,7 +1583,7 @@ ztest_dmu_snapshot_create_destroy(ztest_args_t *za)
 	(void) snprintf(snapname, 100, "%s@%llu", osname,
 	    (u_longlong_t)za->za_instance);
 
-	error = dmu_objset_destroy(snapname);
+	error = dmu_objset_destroy(snapname, B_FALSE);
 	if (error != 0 && error != ENOENT)
 		fatal(0, "dmu_objset_destroy() = %d", error);
 	error = dmu_objset_snapshot(osname, strchr(snapname, '@')+1,
@@ -1614,19 +1614,19 @@ ztest_dsl_dataset_cleanup(char *osname, uint64_t curval)
 	(void) snprintf(clone2name, 100, "%s/c2_%llu", osname, curval);
 	(void) snprintf(snap3name, 100, "%s@s3_%llu", clone1name, curval);
 
-	error = dmu_objset_destroy(clone2name);
+	error = dmu_objset_destroy(clone2name, B_FALSE);
 	if (error && error != ENOENT)
 		fatal(0, "dmu_objset_destroy(%s) = %d", clone2name, error);
-	error = dmu_objset_destroy(snap3name);
+	error = dmu_objset_destroy(snap3name, B_FALSE);
 	if (error && error != ENOENT)
 		fatal(0, "dmu_objset_destroy(%s) = %d", snap3name, error);
-	error = dmu_objset_destroy(snap2name);
+	error = dmu_objset_destroy(snap2name, B_FALSE);
 	if (error && error != ENOENT)
 		fatal(0, "dmu_objset_destroy(%s) = %d", snap2name, error);
-	error = dmu_objset_destroy(clone1name);
+	error = dmu_objset_destroy(clone1name, B_FALSE);
 	if (error && error != ENOENT)
 		fatal(0, "dmu_objset_destroy(%s) = %d", clone1name, error);
-	error = dmu_objset_destroy(snap1name);
+	error = dmu_objset_destroy(snap1name, B_FALSE);
 	if (error && error != ENOENT)
 		fatal(0, "dmu_objset_destroy(%s) = %d", snap1name, error);
 }