Add userland ZFS to vendor/illumos

illumos-gate revision 13742:b6bbdd77139c

cmd/pyzfs/pyzfs.py

@@ -0,0 +1,82 @@
+#! /usr/bin/python2.6 -S
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+# Note, we want SIGINT (control-c) to exit the process quietly, to mimic
+# the standard behavior of C programs.  The best we can do with pure
+# Python is to run with -S (to disable "import site"), and start our
+# program with a "try" statement.  Hopefully nobody hits ^C before our
+# try statement is executed.
+
+try:
+	import site
+	import gettext
+	import zfs.util
+	import zfs.ioctl
+	import sys
+	import errno
+	import solaris.misc
+
+	"""This is the main script for doing zfs subcommands.  It doesn't know
+	what subcommands there are, it just looks for a module zfs.<subcommand>
+	that implements that subcommand."""
+
+	try:
+		_ = gettext.translation("SUNW_OST_OSCMD", "/usr/lib/locale",
+		    fallback=True).gettext
+	except:
+		_ = solaris.misc.gettext
+
+	if len(sys.argv) < 2:
+		sys.exit(_("missing subcommand argument"))
+
+	zfs.ioctl.set_cmdstr(" ".join(["zfs"] + sys.argv[1:]))
+
+	try:
+		# import zfs.<subcommand>
+		# subfunc =  zfs.<subcommand>.do_<subcommand>
+
+		subcmd = sys.argv[1]
+		__import__("zfs." + subcmd)
+		submod = getattr(zfs, subcmd)
+		subfunc = getattr(submod, "do_" + subcmd)
+	except (ImportError, AttributeError):
+		sys.exit(_("invalid subcommand"))
+
+	try:
+		subfunc()
+	except zfs.util.ZFSError, e:
+		print(e)
+		sys.exit(1)
+
+except IOError, e:
+	import errno
+	import sys
+
+	if e.errno == errno.EPIPE:
+		sys.exit(1)
+	raise
+except KeyboardInterrupt:
+	import sys
+
+	sys.exit(1)

cmd/stat/common/statcommon.h

@@ -0,0 +1,328 @@
+/*
+ * 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 (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
+ *
+ * Common routines for acquiring snapshots of kstats for
+ * iostat, mpstat, and vmstat.
+ */
+
+#ifndef	_STATCOMMON_H
+#define	_STATCOMMON_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdio.h>
+#include <kstat.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/buf.h>
+#include <sys/dnlc.h>
+#include <sys/sysinfo.h>
+#include <sys/processor.h>
+#include <sys/pset.h>
+#include <sys/avl.h>
+
+/* No CPU present at this CPU position */
+#define	ID_NO_CPU -1
+/* CPU belongs to no pset (we number this as "pset 0")  */
+#define	ID_NO_PSET 0
+/* CPU is usable */
+#define	CPU_ONLINE(s) ((s) == P_ONLINE || (s) == P_NOINTR)
+/* will the CPU have kstats */
+#define	CPU_ACTIVE(c) (CPU_ONLINE((c)->cs_state) && (c)->cs_id != ID_NO_CPU)
+/* IO device has no identified ID */
+#define	IODEV_NO_ID -1
+/* no limit to iodevs to collect */
+#define	UNLIMITED_IODEVS ((size_t)-1)
+
+#define	NODATE	0	/* Default:  No time stamp */
+#define	DDATE	1	/* Standard date format */
+#define	UDATE	2	/* Internal representation of Unix time */
+
+
+enum snapshot_types {
+	/* All CPUs separately */
+	SNAP_CPUS 		= 1 << 0,
+	/* Aggregated processor sets */
+	SNAP_PSETS		= 1 << 1,
+	/* sys-wide stats including aggregated CPU stats */
+	SNAP_SYSTEM		= 1 << 2,
+	/* interrupt sources and counts */
+	SNAP_INTERRUPTS 	= 1 << 3,
+	/* disk etc. stats */
+	SNAP_IODEVS		= 1 << 4,
+	/* disk controller aggregates */
+	SNAP_CONTROLLERS	= 1 << 5,
+	/* mpxio L I (multipath) paths: -X: Lun,LunInitiator */
+	SNAP_IOPATHS_LI		= 1 << 6,
+	/* mpxio LTI (multipath) paths: -Y: Lun,LunTarget,LunTargetInitiator */
+	SNAP_IOPATHS_LTI	= 1 << 7,
+	/* disk error stats */
+	SNAP_IODEV_ERRORS	= 1 << 8,
+	/* pretty names for iodevs */
+	SNAP_IODEV_PRETTY	= 1 << 9,
+	/* devid for iodevs */
+	SNAP_IODEV_DEVID	= 1 << 10
+};
+
+struct cpu_snapshot {
+	/* may be ID_NO_CPU if no CPU present */
+	processorid_t cs_id;
+	/* may be ID_NO_PSET if no pset */
+	psetid_t cs_pset_id;
+	/* as in p_online(2) */
+	int cs_state;
+	/* stats for this CPU */
+	kstat_t cs_vm;
+	kstat_t cs_sys;
+};
+
+struct pset_snapshot {
+	/* ID may be zero to indicate the "none set" */
+	psetid_t ps_id;
+	/* number of CPUs in set */
+	size_t ps_nr_cpus;
+	/* the CPUs in this set */
+	struct cpu_snapshot **ps_cpus;
+};
+
+struct intr_snapshot {
+	/* name of interrupt source */
+	char is_name[KSTAT_STRLEN];
+	/* total number of interrupts from this source */
+	ulong_t is_total;
+};
+
+struct sys_snapshot {
+	sysinfo_t ss_sysinfo;
+	vminfo_t ss_vminfo;
+	struct nc_stats ss_nc;
+	/* vm/sys stats aggregated across all CPUs */
+	kstat_t ss_agg_vm;
+	kstat_t ss_agg_sys;
+	/* ticks since boot */
+	ulong_t ss_ticks;
+	long ss_deficit;
+};
+
+/* order is significant (see sort_before()) */
+enum iodev_type {
+	IODEV_CONTROLLER	= 1 << 0,
+	IODEV_DISK		= 1 << 1,
+	IODEV_PARTITION		= 1 << 2,
+	IODEV_TAPE		= 1 << 3,
+	IODEV_NFS		= 1 << 4,
+	IODEV_IOPATH_LT		= 1 << 5,	/* synthetic LunTarget */
+	IODEV_IOPATH_LI		= 1 << 6,	/* synthetic LunInitiator */
+	IODEV_IOPATH_LTI	= 1 << 7,	/* LunTgtInitiator (pathinfo) */
+	IODEV_UNKNOWN		= 1 << 8
+};
+
+/* identify a disk, partition, etc. */
+struct iodev_id {
+	int id;
+	/* target id (for disks) */
+	char tid[KSTAT_STRLEN];
+};
+
+/*
+ * Used for disks, partitions, tapes, nfs, controllers, iopaths
+ * Each entry can be a branch of a tree; for example, the disks
+ * of a controller constitute the children of the controller
+ * iodev_snapshot. This relationship is not strictly maintained
+ * if is_pretty can't be found.
+ */
+struct iodev_snapshot {
+	/* original kstat name */
+	char is_name[KSTAT_STRLEN];
+	/* type of kstat */
+	enum iodev_type is_type;
+	/* ID if meaningful */
+	struct iodev_id is_id;
+	/* parent ID if meaningful */
+	struct iodev_id is_parent_id;
+	/* user-friendly name if found */
+	char *is_pretty;
+	/* device ID if applicable */
+	char *is_devid;
+	/* mount-point if applicable */
+	char *is_dname;
+	/* number of direct children */
+	int is_nr_children;
+	/* children of this I/O device */
+	struct iodev_snapshot *is_children;
+	/* standard I/O stats */
+	kstat_io_t is_stats;
+	/* iodev error stats */
+	kstat_t is_errors;
+	/* creation time of the stats */
+	hrtime_t is_crtime;
+	/* time at which iodev snapshot was taken */
+	hrtime_t is_snaptime;
+	/* kstat module */
+	char is_module[KSTAT_STRLEN];
+	/* kstat instance */
+	int is_instance;
+	/* kstat (only used temporarily) */
+	kstat_t *is_ksp;
+	struct iodev_snapshot *is_prev;
+	struct iodev_snapshot *is_next;
+	/* AVL structures to speedup insertion */
+	avl_tree_t *avl_list;	/* list this element belongs to */
+	avl_node_t avl_link;
+};
+
+/* which iodevs to show. */
+struct iodev_filter {
+	/* nr. of iodevs to choose */
+	size_t if_max_iodevs;
+	/* bit mask of enum io_types to allow */
+	int if_allowed_types;
+	/* should we show floppy ? if_names can override this */
+	int if_skip_floppy;
+	/* nr. of named iodevs */
+	size_t if_nr_names;
+	char **if_names;
+};
+
+/* The primary structure of a system snapshot. */
+struct snapshot {
+	/* what types were *requested* */
+	enum snapshot_types s_types;
+	size_t s_nr_cpus;
+	struct cpu_snapshot *s_cpus;
+	size_t s_nr_psets;
+	struct pset_snapshot *s_psets;
+	size_t s_nr_intrs;
+	struct intr_snapshot *s_intrs;
+	size_t s_nr_iodevs;
+	struct iodev_snapshot *s_iodevs;
+	size_t s_iodevs_is_name_maxlen;
+	struct sys_snapshot s_sys;
+	struct biostats s_biostats;
+};
+
+/* print a message and exit with failure */
+void fail(int do_perror, char *message, ...);
+
+/* strdup str, or exit with failure */
+char *safe_strdup(char *str);
+
+/* malloc successfully, or exit with failure */
+void *safe_alloc(size_t size);
+
+/*
+ * Copy a kstat from src to dst. If the source kstat contains no data,
+ * then set the destination kstat data to NULL and size to zero.
+ * Returns 0 on success.
+ */
+int kstat_copy(const kstat_t *src, kstat_t *dst);
+
+/*
+ * Look up the named kstat, and give the ui64 difference i.e.
+ * new - old, or if old is NULL, return new.
+ */
+uint64_t kstat_delta(kstat_t *old, kstat_t *new, char *name);
+
+/* Return the number of ticks delta between two hrtime_t values. */
+uint64_t hrtime_delta(hrtime_t old, hrtime_t new);
+
+/*
+ * Add the integer-valued stats from "src" to the
+ * existing ones in "dst". If "dst" does not contain
+ * stats, then a kstat_copy() is performed.
+ */
+int kstat_add(const kstat_t *src, kstat_t *dst);
+
+/* return the number of CPUs with kstats (i.e. present and online) */
+int nr_active_cpus(struct snapshot *ss);
+
+/*
+ * Return the difference in CPU ticks between the two sys
+ * kstats.
+ */
+uint64_t cpu_ticks_delta(kstat_t *old, kstat_t *new);
+
+/*
+ * Open the kstat chain. Cannot fail.
+ */
+kstat_ctl_t *open_kstat(void);
+
+/*
+ * Return a struct snapshot based on the snapshot_types parameter
+ * passed in. iodev_filter may be NULL in which case all iodevs
+ * are selected if SNAP_IODEVS is passed.
+ */
+struct snapshot *acquire_snapshot(kstat_ctl_t *, int, struct iodev_filter *);
+
+/* free a snapshot */
+void free_snapshot(struct snapshot *ss);
+
+typedef void (*snapshot_cb)(void *old, void *new, void *data);
+
+/*
+ * Call the call back for each pair of data items of the given type,
+ * passing the data pointer passed in as well. If an item has been
+ * added, the first pointer will be NULL; if removed, the second pointer
+ * will be NULL.
+ *
+ * A non-zero return value indicates configuration has changed.
+ */
+int snapshot_walk(enum snapshot_types type, struct snapshot *old,
+    struct snapshot *new, snapshot_cb cb, void *data);
+
+/*
+ * Output a line detailing any configuration changes such as a CPU
+ * brought online, etc, bracketed by << >>.
+ */
+void snapshot_report_changes(struct snapshot *old, struct snapshot *new);
+
+/* Return non-zero if configuration has changed. */
+int snapshot_has_changed(struct snapshot *old, struct snapshot *new);
+
+/* free the given iodev */
+void free_iodev(struct iodev_snapshot *iodev);
+
+/* acquire the I/O devices */
+int acquire_iodevs(struct snapshot *ss, kstat_ctl_t *kc,
+    struct iodev_filter *df);
+
+/* strcmp-style I/O device comparator */
+int iodev_cmp(struct iodev_snapshot *io1, struct iodev_snapshot *io2);
+
+/* sleep until *wakeup + interval, keeping cadence where desired */
+void sleep_until(hrtime_t *wakeup, hrtime_t interval, int forever,
+    int *caught_cont);
+
+/* signal handler - so we can be aware of SIGCONT */
+void cont_handler(int sig_number);
+
+/* Print a timestamp in either Unix or standard format. */
+void print_timestamp(uint_t);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _STATCOMMON_H */

cmd/stat/common/timestamp.c

@@ -0,0 +1,54 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include "statcommon.h"
+
+#include <langinfo.h>
+
+/*
+ * Print timestamp as decimal reprentation of time_t value (-T u was specified)
+ * or in date(1) format (-T d was specified).
+ */
+void
+print_timestamp(uint_t timestamp_fmt)
+{
+	time_t t = time(NULL);
+	static char *fmt = NULL;
+
+	/* We only need to retrieve this once per invocation */
+	if (fmt == NULL)
+		fmt = nl_langinfo(_DATE_FMT);
+
+	if (timestamp_fmt == UDATE) {
+		(void) printf("%ld\n", t);
+	} else if (timestamp_fmt == DDATE) {
+		char dstr[64];
+		int len;
+
+		len = strftime(dstr, sizeof (dstr), fmt, localtime(&t));
+		if (len > 0)
+			(void) printf("%s\n", dstr);
+	}
+}

cmd/zdb/zdb_il.c

@@ -0,0 +1,384 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+/*
+ * Print intent log header and statistics.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <ctype.h>
+#include <sys/zfs_context.h>
+#include <sys/spa.h>
+#include <sys/dmu.h>
+#include <sys/stat.h>
+#include <sys/resource.h>
+#include <sys/zil.h>
+#include <sys/zil_impl.h>
+
+extern uint8_t dump_opt[256];
+
+static char prefix[4] = "\t\t\t";
+
+static void
+print_log_bp(const blkptr_t *bp, const char *prefix)
+{
+	char blkbuf[BP_SPRINTF_LEN];
+
+	sprintf_blkptr(blkbuf, bp);
+	(void) printf("%s%s\n", prefix, blkbuf);
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_create(zilog_t *zilog, int txtype, lr_create_t *lr)
+{
+	time_t crtime = lr->lr_crtime[0];
+	char *name, *link;
+	lr_attr_t *lrattr;
+
+	name = (char *)(lr + 1);
+
+	if (lr->lr_common.lrc_txtype == TX_CREATE_ATTR ||
+	    lr->lr_common.lrc_txtype == TX_MKDIR_ATTR) {
+		lrattr = (lr_attr_t *)(lr + 1);
+		name += ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
+	}
+
+	if (txtype == TX_SYMLINK) {
+		link = name + strlen(name) + 1;
+		(void) printf("%s%s -> %s\n", prefix, name, link);
+	} else if (txtype != TX_MKXATTR) {
+		(void) printf("%s%s\n", prefix, name);
+	}
+
+	(void) printf("%s%s", prefix, ctime(&crtime));
+	(void) printf("%sdoid %llu, foid %llu, mode %llo\n", prefix,
+	    (u_longlong_t)lr->lr_doid, (u_longlong_t)lr->lr_foid,
+	    (longlong_t)lr->lr_mode);
+	(void) printf("%suid %llu, gid %llu, gen %llu, rdev 0x%llx\n", prefix,
+	    (u_longlong_t)lr->lr_uid, (u_longlong_t)lr->lr_gid,
+	    (u_longlong_t)lr->lr_gen, (u_longlong_t)lr->lr_rdev);
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_remove(zilog_t *zilog, int txtype, lr_remove_t *lr)
+{
+	(void) printf("%sdoid %llu, name %s\n", prefix,
+	    (u_longlong_t)lr->lr_doid, (char *)(lr + 1));
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_link(zilog_t *zilog, int txtype, lr_link_t *lr)
+{
+	(void) printf("%sdoid %llu, link_obj %llu, name %s\n", prefix,
+	    (u_longlong_t)lr->lr_doid, (u_longlong_t)lr->lr_link_obj,
+	    (char *)(lr + 1));
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_rename(zilog_t *zilog, int txtype, lr_rename_t *lr)
+{
+	char *snm = (char *)(lr + 1);
+	char *tnm = snm + strlen(snm) + 1;
+
+	(void) printf("%ssdoid %llu, tdoid %llu\n", prefix,
+	    (u_longlong_t)lr->lr_sdoid, (u_longlong_t)lr->lr_tdoid);
+	(void) printf("%ssrc %s tgt %s\n", prefix, snm, tnm);
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_write(zilog_t *zilog, int txtype, lr_write_t *lr)
+{
+	char *data, *dlimit;
+	blkptr_t *bp = &lr->lr_blkptr;
+	zbookmark_t zb;
+	char buf[SPA_MAXBLOCKSIZE];
+	int verbose = MAX(dump_opt['d'], dump_opt['i']);
+	int error;
+
+	(void) printf("%sfoid %llu, offset %llx, length %llx\n", prefix,
+	    (u_longlong_t)lr->lr_foid, (u_longlong_t)lr->lr_offset,
+	    (u_longlong_t)lr->lr_length);
+
+	if (txtype == TX_WRITE2 || verbose < 5)
+		return;
+
+	if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) {
+		(void) printf("%shas blkptr, %s\n", prefix,
+		    bp->blk_birth >= spa_first_txg(zilog->zl_spa) ?
+		    "will claim" : "won't claim");
+		print_log_bp(bp, prefix);
+
+		if (BP_IS_HOLE(bp)) {
+			(void) printf("\t\t\tLSIZE 0x%llx\n",
+			    (u_longlong_t)BP_GET_LSIZE(bp));
+		}
+		if (bp->blk_birth == 0) {
+			bzero(buf, sizeof (buf));
+			(void) printf("%s<hole>\n", prefix);
+			return;
+		}
+		if (bp->blk_birth < zilog->zl_header->zh_claim_txg) {
+			(void) printf("%s<block already committed>\n", prefix);
+			return;
+		}
+
+		SET_BOOKMARK(&zb, dmu_objset_id(zilog->zl_os),
+		    lr->lr_foid, ZB_ZIL_LEVEL,
+		    lr->lr_offset / BP_GET_LSIZE(bp));
+
+		error = zio_wait(zio_read(NULL, zilog->zl_spa,
+		    bp, buf, BP_GET_LSIZE(bp), NULL, NULL,
+		    ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &zb));
+		if (error)
+			return;
+		data = buf;
+	} else {
+		data = (char *)(lr + 1);
+	}
+
+	dlimit = data + MIN(lr->lr_length,
+	    (verbose < 6 ? 20 : SPA_MAXBLOCKSIZE));
+
+	(void) printf("%s", prefix);
+	while (data < dlimit) {
+		if (isprint(*data))
+			(void) printf("%c ", *data);
+		else
+			(void) printf("%2X", *data);
+		data++;
+	}
+	(void) printf("\n");
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_truncate(zilog_t *zilog, int txtype, lr_truncate_t *lr)
+{
+	(void) printf("%sfoid %llu, offset 0x%llx, length 0x%llx\n", prefix,
+	    (u_longlong_t)lr->lr_foid, (longlong_t)lr->lr_offset,
+	    (u_longlong_t)lr->lr_length);
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_setattr(zilog_t *zilog, int txtype, lr_setattr_t *lr)
+{
+	time_t atime = (time_t)lr->lr_atime[0];
+	time_t mtime = (time_t)lr->lr_mtime[0];
+
+	(void) printf("%sfoid %llu, mask 0x%llx\n", prefix,
+	    (u_longlong_t)lr->lr_foid, (u_longlong_t)lr->lr_mask);
+
+	if (lr->lr_mask & AT_MODE) {
+		(void) printf("%sAT_MODE  %llo\n", prefix,
+		    (longlong_t)lr->lr_mode);
+	}
+
+	if (lr->lr_mask & AT_UID) {
+		(void) printf("%sAT_UID   %llu\n", prefix,
+		    (u_longlong_t)lr->lr_uid);
+	}
+
+	if (lr->lr_mask & AT_GID) {
+		(void) printf("%sAT_GID   %llu\n", prefix,
+		    (u_longlong_t)lr->lr_gid);
+	}
+
+	if (lr->lr_mask & AT_SIZE) {
+		(void) printf("%sAT_SIZE  %llu\n", prefix,
+		    (u_longlong_t)lr->lr_size);
+	}
+
+	if (lr->lr_mask & AT_ATIME) {
+		(void) printf("%sAT_ATIME %llu.%09llu %s", prefix,
+		    (u_longlong_t)lr->lr_atime[0],
+		    (u_longlong_t)lr->lr_atime[1],
+		    ctime(&atime));
+	}
+
+	if (lr->lr_mask & AT_MTIME) {
+		(void) printf("%sAT_MTIME %llu.%09llu %s", prefix,
+		    (u_longlong_t)lr->lr_mtime[0],
+		    (u_longlong_t)lr->lr_mtime[1],
+		    ctime(&mtime));
+	}
+}
+
+/* ARGSUSED */
+static void
+zil_prt_rec_acl(zilog_t *zilog, int txtype, lr_acl_t *lr)
+{
+	(void) printf("%sfoid %llu, aclcnt %llu\n", prefix,
+	    (u_longlong_t)lr->lr_foid, (u_longlong_t)lr->lr_aclcnt);
+}
+
+typedef void (*zil_prt_rec_func_t)();
+typedef struct zil_rec_info {
+	zil_prt_rec_func_t	zri_print;
+	char			*zri_name;
+	uint64_t		zri_count;
+} zil_rec_info_t;
+
+static zil_rec_info_t zil_rec_info[TX_MAX_TYPE] = {
+	{	NULL,			"Total              " },
+	{	zil_prt_rec_create,	"TX_CREATE          " },
+	{	zil_prt_rec_create,	"TX_MKDIR           " },
+	{	zil_prt_rec_create,	"TX_MKXATTR         " },
+	{	zil_prt_rec_create,	"TX_SYMLINK         " },
+	{	zil_prt_rec_remove,	"TX_REMOVE          " },
+	{	zil_prt_rec_remove,	"TX_RMDIR           " },
+	{	zil_prt_rec_link,	"TX_LINK            " },
+	{	zil_prt_rec_rename,	"TX_RENAME          " },
+	{	zil_prt_rec_write,	"TX_WRITE           " },
+	{	zil_prt_rec_truncate,	"TX_TRUNCATE        " },
+	{	zil_prt_rec_setattr,	"TX_SETATTR         " },
+	{	zil_prt_rec_acl,	"TX_ACL_V0          " },
+	{	zil_prt_rec_acl,	"TX_ACL_ACL         " },
+	{	zil_prt_rec_create,	"TX_CREATE_ACL      " },
+	{	zil_prt_rec_create,	"TX_CREATE_ATTR     " },
+	{	zil_prt_rec_create,	"TX_CREATE_ACL_ATTR " },
+	{	zil_prt_rec_create,	"TX_MKDIR_ACL       " },
+	{	zil_prt_rec_create,	"TX_MKDIR_ATTR      " },
+	{	zil_prt_rec_create,	"TX_MKDIR_ACL_ATTR  " },
+	{	zil_prt_rec_write,	"TX_WRITE2          " },
+};
+
+/* ARGSUSED */
+static int
+print_log_record(zilog_t *zilog, lr_t *lr, void *arg, uint64_t claim_txg)
+{
+	int txtype;
+	int verbose = MAX(dump_opt['d'], dump_opt['i']);
+
+	/* reduce size of txtype to strip off TX_CI bit */
+	txtype = lr->lrc_txtype;
+
+	ASSERT(txtype != 0 && (uint_t)txtype < TX_MAX_TYPE);
+	ASSERT(lr->lrc_txg);
+
+	(void) printf("\t\t%s%s len %6llu, txg %llu, seq %llu\n",
+	    (lr->lrc_txtype & TX_CI) ? "CI-" : "",
+	    zil_rec_info[txtype].zri_name,
+	    (u_longlong_t)lr->lrc_reclen,
+	    (u_longlong_t)lr->lrc_txg,
+	    (u_longlong_t)lr->lrc_seq);
+
+	if (txtype && verbose >= 3)
+		zil_rec_info[txtype].zri_print(zilog, txtype, lr);
+
+	zil_rec_info[txtype].zri_count++;
+	zil_rec_info[0].zri_count++;
+
+	return (0);
+}
+
+/* ARGSUSED */
+static int
+print_log_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg)
+{
+	char blkbuf[BP_SPRINTF_LEN + 10];
+	int verbose = MAX(dump_opt['d'], dump_opt['i']);
+	char *claim;
+
+	if (verbose <= 3)
+		return (0);
+
+	if (verbose >= 5) {
+		(void) strcpy(blkbuf, ", ");
+		sprintf_blkptr(blkbuf + strlen(blkbuf), bp);
+	} else {
+		blkbuf[0] = '\0';
+	}
+
+	if (claim_txg != 0)
+		claim = "already claimed";
+	else if (bp->blk_birth >= spa_first_txg(zilog->zl_spa))
+		claim = "will claim";
+	else
+		claim = "won't claim";
+
+	(void) printf("\tBlock seqno %llu, %s%s\n",
+	    (u_longlong_t)bp->blk_cksum.zc_word[ZIL_ZC_SEQ], claim, blkbuf);
+
+	return (0);
+}
+
+static void
+print_log_stats(int verbose)
+{
+	int i, w, p10;
+
+	if (verbose > 3)
+		(void) printf("\n");
+
+	if (zil_rec_info[0].zri_count == 0)
+		return;
+
+	for (w = 1, p10 = 10; zil_rec_info[0].zri_count >= p10; p10 *= 10)
+		w++;
+
+	for (i = 0; i < TX_MAX_TYPE; i++)
+		if (zil_rec_info[i].zri_count || verbose >= 3)
+			(void) printf("\t\t%s %*llu\n",
+			    zil_rec_info[i].zri_name, w,
+			    (u_longlong_t)zil_rec_info[i].zri_count);
+	(void) printf("\n");
+}
+
+/* ARGSUSED */
+void
+dump_intent_log(zilog_t *zilog)
+{
+	const zil_header_t *zh = zilog->zl_header;
+	int verbose = MAX(dump_opt['d'], dump_opt['i']);
+	int i;
+
+	if (zh->zh_log.blk_birth == 0 || verbose < 1)
+		return;
+
+	(void) printf("\n    ZIL header: claim_txg %llu, "
+	    "claim_blk_seq %llu, claim_lr_seq %llu",
+	    (u_longlong_t)zh->zh_claim_txg,
+	    (u_longlong_t)zh->zh_claim_blk_seq,
+	    (u_longlong_t)zh->zh_claim_lr_seq);
+	(void) printf(" replay_seq %llu, flags 0x%llx\n",
+	    (u_longlong_t)zh->zh_replay_seq, (u_longlong_t)zh->zh_flags);
+
+	for (i = 0; i < TX_MAX_TYPE; i++)
+		zil_rec_info[i].zri_count = 0;
+
+	if (verbose >= 2) {
+		(void) printf("\n");
+		(void) zil_parse(zilog, print_log_block, print_log_record, NULL,
+		    zh->zh_claim_txg);
+		print_log_stats(verbose);
+	}
+}

cmd/zfs/zfs_iter.c

@@ -0,0 +1,464 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#include <libintl.h>
+#include <libuutil.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+
+#include <libzfs.h>
+
+#include "zfs_util.h"
+#include "zfs_iter.h"
+
+/*
+ * This is a private interface used to gather up all the datasets specified on
+ * the command line so that we can iterate over them in order.
+ *
+ * First, we iterate over all filesystems, gathering them together into an
+ * AVL tree.  We report errors for any explicitly specified datasets
+ * that we couldn't open.
+ *
+ * When finished, we have an AVL tree of ZFS handles.  We go through and execute
+ * the provided callback for each one, passing whatever data the user supplied.
+ */
+
+typedef struct zfs_node {
+	zfs_handle_t	*zn_handle;
+	uu_avl_node_t	zn_avlnode;
+} zfs_node_t;
+
+typedef struct callback_data {
+	uu_avl_t		*cb_avl;
+	int			cb_flags;
+	zfs_type_t		cb_types;
+	zfs_sort_column_t	*cb_sortcol;
+	zprop_list_t		**cb_proplist;
+	int			cb_depth_limit;
+	int			cb_depth;
+	uint8_t			cb_props_table[ZFS_NUM_PROPS];
+} callback_data_t;
+
+uu_avl_pool_t *avl_pool;
+
+/*
+ * Include snaps if they were requested or if this a zfs list where types
+ * were not specified and the "listsnapshots" property is set on this pool.
+ */
+static int
+zfs_include_snapshots(zfs_handle_t *zhp, callback_data_t *cb)
+{
+	zpool_handle_t *zph;
+
+	if ((cb->cb_flags & ZFS_ITER_PROP_LISTSNAPS) == 0)
+		return (cb->cb_types & ZFS_TYPE_SNAPSHOT);
+
+	zph = zfs_get_pool_handle(zhp);
+	return (zpool_get_prop_int(zph, ZPOOL_PROP_LISTSNAPS, NULL));
+}
+
+/*
+ * Called for each dataset.  If the object is of an appropriate type,
+ * add it to the avl tree and recurse over any children as necessary.
+ */
+static int
+zfs_callback(zfs_handle_t *zhp, void *data)
+{
+	callback_data_t *cb = data;
+	int dontclose = 0;
+	int include_snaps = zfs_include_snapshots(zhp, cb);
+
+	if ((zfs_get_type(zhp) & cb->cb_types) ||
+	    ((zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT) && include_snaps)) {
+		uu_avl_index_t idx;
+		zfs_node_t *node = safe_malloc(sizeof (zfs_node_t));
+
+		node->zn_handle = zhp;
+		uu_avl_node_init(node, &node->zn_avlnode, avl_pool);
+		if (uu_avl_find(cb->cb_avl, node, cb->cb_sortcol,
+		    &idx) == NULL) {
+			if (cb->cb_proplist) {
+				if ((*cb->cb_proplist) &&
+				    !(*cb->cb_proplist)->pl_all)
+					zfs_prune_proplist(zhp,
+					    cb->cb_props_table);
+
+				if (zfs_expand_proplist(zhp, cb->cb_proplist,
+				    (cb->cb_flags & ZFS_ITER_RECVD_PROPS))
+				    != 0) {
+					free(node);
+					return (-1);
+				}
+			}
+			uu_avl_insert(cb->cb_avl, node, idx);
+			dontclose = 1;
+		} else {
+			free(node);
+		}
+	}
+
+	/*
+	 * Recurse if necessary.
+	 */
+	if (cb->cb_flags & ZFS_ITER_RECURSE &&
+	    ((cb->cb_flags & ZFS_ITER_DEPTH_LIMIT) == 0 ||
+	    cb->cb_depth < cb->cb_depth_limit)) {
+		cb->cb_depth++;
+		if (zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM)
+			(void) zfs_iter_filesystems(zhp, zfs_callback, data);
+		if ((zfs_get_type(zhp) != ZFS_TYPE_SNAPSHOT) && include_snaps)
+			(void) zfs_iter_snapshots(zhp, zfs_callback, data);
+		cb->cb_depth--;
+	}
+
+	if (!dontclose)
+		zfs_close(zhp);
+
+	return (0);
+}
+
+int
+zfs_add_sort_column(zfs_sort_column_t **sc, const char *name,
+    boolean_t reverse)
+{
+	zfs_sort_column_t *col;
+	zfs_prop_t prop;
+
+	if ((prop = zfs_name_to_prop(name)) == ZPROP_INVAL &&
+	    !zfs_prop_user(name))
+		return (-1);
+
+	col = safe_malloc(sizeof (zfs_sort_column_t));
+
+	col->sc_prop = prop;
+	col->sc_reverse = reverse;
+	if (prop == ZPROP_INVAL) {
+		col->sc_user_prop = safe_malloc(strlen(name) + 1);
+		(void) strcpy(col->sc_user_prop, name);
+	}
+
+	if (*sc == NULL) {
+		col->sc_last = col;
+		*sc = col;
+	} else {
+		(*sc)->sc_last->sc_next = col;
+		(*sc)->sc_last = col;
+	}
+
+	return (0);
+}
+
+void
+zfs_free_sort_columns(zfs_sort_column_t *sc)
+{
+	zfs_sort_column_t *col;
+
+	while (sc != NULL) {
+		col = sc->sc_next;
+		free(sc->sc_user_prop);
+		free(sc);
+		sc = col;
+	}
+}
+
+/* ARGSUSED */
+static int
+zfs_compare(const void *larg, const void *rarg, void *unused)
+{
+	zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle;
+	zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle;
+	const char *lname = zfs_get_name(l);
+	const char *rname = zfs_get_name(r);
+	char *lat, *rat;
+	uint64_t lcreate, rcreate;
+	int ret;
+
+	lat = (char *)strchr(lname, '@');
+	rat = (char *)strchr(rname, '@');
+
+	if (lat != NULL)
+		*lat = '\0';
+	if (rat != NULL)
+		*rat = '\0';
+
+	ret = strcmp(lname, rname);
+	if (ret == 0) {
+		/*
+		 * If we're comparing a dataset to one of its snapshots, we
+		 * always make the full dataset first.
+		 */
+		if (lat == NULL) {
+			ret = -1;
+		} else if (rat == NULL) {
+			ret = 1;
+		} else {
+			/*
+			 * If we have two snapshots from the same dataset, then
+			 * we want to sort them according to creation time.  We
+			 * use the hidden CREATETXG property to get an absolute
+			 * ordering of snapshots.
+			 */
+			lcreate = zfs_prop_get_int(l, ZFS_PROP_CREATETXG);
+			rcreate = zfs_prop_get_int(r, ZFS_PROP_CREATETXG);
+
+			if (lcreate < rcreate)
+				ret = -1;
+			else if (lcreate > rcreate)
+				ret = 1;
+		}
+	}
+
+	if (lat != NULL)
+		*lat = '@';
+	if (rat != NULL)
+		*rat = '@';
+
+	return (ret);
+}
+
+/*
+ * Sort datasets by specified columns.
+ *
+ * o  Numeric types sort in ascending order.
+ * o  String types sort in alphabetical order.
+ * o  Types inappropriate for a row sort that row to the literal
+ *    bottom, regardless of the specified ordering.
+ *
+ * If no sort columns are specified, or two datasets compare equally
+ * across all specified columns, they are sorted alphabetically by name
+ * with snapshots grouped under their parents.
+ */
+static int
+zfs_sort(const void *larg, const void *rarg, void *data)
+{
+	zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle;
+	zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle;
+	zfs_sort_column_t *sc = (zfs_sort_column_t *)data;
+	zfs_sort_column_t *psc;
+
+	for (psc = sc; psc != NULL; psc = psc->sc_next) {
+		char lbuf[ZFS_MAXPROPLEN], rbuf[ZFS_MAXPROPLEN];
+		char *lstr, *rstr;
+		uint64_t lnum, rnum;
+		boolean_t lvalid, rvalid;
+		int ret = 0;
+
+		/*
+		 * We group the checks below the generic code.  If 'lstr' and
+		 * 'rstr' are non-NULL, then we do a string based comparison.
+		 * Otherwise, we compare 'lnum' and 'rnum'.
+		 */
+		lstr = rstr = NULL;
+		if (psc->sc_prop == ZPROP_INVAL) {
+			nvlist_t *luser, *ruser;
+			nvlist_t *lval, *rval;
+
+			luser = zfs_get_user_props(l);
+			ruser = zfs_get_user_props(r);
+
+			lvalid = (nvlist_lookup_nvlist(luser,
+			    psc->sc_user_prop, &lval) == 0);
+			rvalid = (nvlist_lookup_nvlist(ruser,
+			    psc->sc_user_prop, &rval) == 0);
+
+			if (lvalid)
+				verify(nvlist_lookup_string(lval,
+				    ZPROP_VALUE, &lstr) == 0);
+			if (rvalid)
+				verify(nvlist_lookup_string(rval,
+				    ZPROP_VALUE, &rstr) == 0);
+
+		} else if (zfs_prop_is_string(psc->sc_prop)) {
+			lvalid = (zfs_prop_get(l, psc->sc_prop, lbuf,
+			    sizeof (lbuf), NULL, NULL, 0, B_TRUE) == 0);
+			rvalid = (zfs_prop_get(r, psc->sc_prop, rbuf,
+			    sizeof (rbuf), NULL, NULL, 0, B_TRUE) == 0);
+
+			lstr = lbuf;
+			rstr = rbuf;
+		} else {
+			lvalid = zfs_prop_valid_for_type(psc->sc_prop,
+			    zfs_get_type(l));
+			rvalid = zfs_prop_valid_for_type(psc->sc_prop,
+			    zfs_get_type(r));
+
+			if (lvalid)
+				(void) zfs_prop_get_numeric(l, psc->sc_prop,
+				    &lnum, NULL, NULL, 0);
+			if (rvalid)
+				(void) zfs_prop_get_numeric(r, psc->sc_prop,
+				    &rnum, NULL, NULL, 0);
+		}
+
+		if (!lvalid && !rvalid)
+			continue;
+		else if (!lvalid)
+			return (1);
+		else if (!rvalid)
+			return (-1);
+
+		if (lstr)
+			ret = strcmp(lstr, rstr);
+		else if (lnum < rnum)
+			ret = -1;
+		else if (lnum > rnum)
+			ret = 1;
+
+		if (ret != 0) {
+			if (psc->sc_reverse == B_TRUE)
+				ret = (ret < 0) ? 1 : -1;
+			return (ret);
+		}
+	}
+
+	return (zfs_compare(larg, rarg, NULL));
+}
+
+int
+zfs_for_each(int argc, char **argv, int flags, zfs_type_t types,
+    zfs_sort_column_t *sortcol, zprop_list_t **proplist, int limit,
+    zfs_iter_f callback, void *data)
+{
+	callback_data_t cb = {0};
+	int ret = 0;
+	zfs_node_t *node;
+	uu_avl_walk_t *walk;
+
+	avl_pool = uu_avl_pool_create("zfs_pool", sizeof (zfs_node_t),
+	    offsetof(zfs_node_t, zn_avlnode), zfs_sort, UU_DEFAULT);
+
+	if (avl_pool == NULL)
+		nomem();
+
+	cb.cb_sortcol = sortcol;
+	cb.cb_flags = flags;
+	cb.cb_proplist = proplist;
+	cb.cb_types = types;
+	cb.cb_depth_limit = limit;
+	/*
+	 * 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
+	 * accessed.
+	 *
+	 * If cb_proplist is NULL then we retain all the properties.  We
+	 * always retain the zoned property, which some other properties
+	 * need (userquota & friends), and the createtxg property, which
+	 * we need to sort snapshots.
+	 */
+	if (cb.cb_proplist && *cb.cb_proplist) {
+		zprop_list_t *p = *cb.cb_proplist;
+
+		while (p) {
+			if (p->pl_prop >= ZFS_PROP_TYPE &&
+			    p->pl_prop < ZFS_NUM_PROPS) {
+				cb.cb_props_table[p->pl_prop] = B_TRUE;
+			}
+			p = p->pl_next;
+		}
+
+		while (sortcol) {
+			if (sortcol->sc_prop >= ZFS_PROP_TYPE &&
+			    sortcol->sc_prop < ZFS_NUM_PROPS) {
+				cb.cb_props_table[sortcol->sc_prop] = B_TRUE;
+			}
+			sortcol = sortcol->sc_next;
+		}
+
+		cb.cb_props_table[ZFS_PROP_ZONED] = B_TRUE;
+		cb.cb_props_table[ZFS_PROP_CREATETXG] = B_TRUE;
+	} else {
+		(void) memset(cb.cb_props_table, B_TRUE,
+		    sizeof (cb.cb_props_table));
+	}
+
+	if ((cb.cb_avl = uu_avl_create(avl_pool, NULL, UU_DEFAULT)) == NULL)
+		nomem();
+
+	if (argc == 0) {
+		/*
+		 * If given no arguments, iterate over all datasets.
+		 */
+		cb.cb_flags |= ZFS_ITER_RECURSE;
+		ret = zfs_iter_root(g_zfs, zfs_callback, &cb);
+	} else {
+		int i;
+		zfs_handle_t *zhp;
+		zfs_type_t argtype;
+
+		/*
+		 * If we're recursive, then we always allow filesystems as
+		 * arguments.  If we also are interested in snapshots, then we
+		 * can take volumes as well.
+		 */
+		argtype = types;
+		if (flags & ZFS_ITER_RECURSE) {
+			argtype |= ZFS_TYPE_FILESYSTEM;
+			if (types & ZFS_TYPE_SNAPSHOT)
+				argtype |= ZFS_TYPE_VOLUME;
+		}
+
+		for (i = 0; i < argc; i++) {
+			if (flags & ZFS_ITER_ARGS_CAN_BE_PATHS) {
+				zhp = zfs_path_to_zhandle(g_zfs, argv[i],
+				    argtype);
+			} else {
+				zhp = zfs_open(g_zfs, argv[i], argtype);
+			}
+			if (zhp != NULL)
+				ret |= zfs_callback(zhp, &cb);
+			else
+				ret = 1;
+		}
+	}
+
+	/*
+	 * At this point we've got our AVL tree full of zfs handles, so iterate
+	 * over each one and execute the real user callback.
+	 */
+	for (node = uu_avl_first(cb.cb_avl); node != NULL;
+	    node = uu_avl_next(cb.cb_avl, node))
+		ret |= callback(node->zn_handle, data);
+
+	/*
+	 * Finally, clean up the AVL tree.
+	 */
+	if ((walk = uu_avl_walk_start(cb.cb_avl, UU_WALK_ROBUST)) == NULL)
+		nomem();
+
+	while ((node = uu_avl_walk_next(walk)) != NULL) {
+		uu_avl_remove(cb.cb_avl, node);
+		zfs_close(node->zn_handle);
+		free(node);
+	}
+
+	uu_avl_walk_end(walk);
+	uu_avl_destroy(cb.cb_avl);
+	uu_avl_pool_destroy(avl_pool);
+
+	return (ret);
+}

cmd/zfs/zfs_iter.h

@@ -0,0 +1,56 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef	ZFS_ITER_H
+#define	ZFS_ITER_H
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+typedef struct zfs_sort_column {
+	struct zfs_sort_column	*sc_next;
+	struct zfs_sort_column	*sc_last;
+	zfs_prop_t		sc_prop;
+	char			*sc_user_prop;
+	boolean_t		sc_reverse;
+} zfs_sort_column_t;
+
+#define	ZFS_ITER_RECURSE	   (1 << 0)
+#define	ZFS_ITER_ARGS_CAN_BE_PATHS (1 << 1)
+#define	ZFS_ITER_PROP_LISTSNAPS    (1 << 2)
+#define	ZFS_ITER_DEPTH_LIMIT	   (1 << 3)
+#define	ZFS_ITER_RECVD_PROPS	   (1 << 4)
+
+int zfs_for_each(int, char **, int options, zfs_type_t,
+    zfs_sort_column_t *, zprop_list_t **, int, zfs_iter_f, void *);
+int zfs_add_sort_column(zfs_sort_column_t **, const char *, boolean_t);
+void zfs_free_sort_columns(zfs_sort_column_t *);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* ZFS_ITER_H */

cmd/zfs/zfs_util.h

@@ -0,0 +1,42 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef	_ZFS_UTIL_H
+#define	_ZFS_UTIL_H
+
+#include <libzfs.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+void * safe_malloc(size_t size);
+void nomem(void);
+libzfs_handle_t *g_zfs;
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _ZFS_UTIL_H */

cmd/zhack/zhack.c

@@ -0,0 +1,533 @@
+/*
+ * 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 (c) 2012 by Delphix. All rights reserved.
+ */
+
+/*
+ * zhack is a debugging tool that can write changes to ZFS pool using libzpool
+ * for testing purposes. Altering pools with zhack is unsupported and may
+ * result in corrupted pools.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <ctype.h>
+#include <sys/zfs_context.h>
+#include <sys/spa.h>
+#include <sys/spa_impl.h>
+#include <sys/dmu.h>
+#include <sys/zap.h>
+#include <sys/zfs_znode.h>
+#include <sys/dsl_synctask.h>
+#include <sys/vdev.h>
+#include <sys/fs/zfs.h>
+#include <sys/dmu_objset.h>
+#include <sys/dsl_pool.h>
+#include <sys/zio_checksum.h>
+#include <sys/zio_compress.h>
+#include <sys/zfeature.h>
+#undef ZFS_MAXNAMELEN
+#undef verify
+#include <libzfs.h>
+
+extern boolean_t zfeature_checks_disable;
+
+const char cmdname[] = "zhack";
+libzfs_handle_t *g_zfs;
+static importargs_t g_importargs;
+static char *g_pool;
+static boolean_t g_readonly;
+
+static void
+usage(void)
+{
+	(void) fprintf(stderr,
+	    "Usage: %s [-c cachefile] [-d dir] <subcommand> <args> ...\n"
+	    "where <subcommand> <args> is one of the following:\n"
+	    "\n", cmdname);
+
+	(void) fprintf(stderr,
+	    "    feature stat <pool>\n"
+	    "        print information about enabled features\n"
+	    "    feature enable [-d desc] <pool> <feature>\n"
+	    "        add a new enabled feature to the pool\n"
+	    "        -d <desc> sets the feature's description\n"
+	    "    feature ref [-md] <pool> <feature>\n"
+	    "        change the refcount on the given feature\n"
+	    "        -d decrease instead of increase the refcount\n"
+	    "        -m add the feature to the label if increasing refcount\n"
+	    "\n"
+	    "    <feature> : should be a feature guid\n");
+	exit(1);
+}
+
+
+static void
+fatal(const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	(void) fprintf(stderr, "%s: ", cmdname);
+	(void) vfprintf(stderr, fmt, ap);
+	va_end(ap);
+	(void) fprintf(stderr, "\n");
+
+	exit(1);
+}
+
+/* ARGSUSED */
+static int
+space_delta_cb(dmu_object_type_t bonustype, void *data,
+    uint64_t *userp, uint64_t *groupp)
+{
+	/*
+	 * Is it a valid type of object to track?
+	 */
+	if (bonustype != DMU_OT_ZNODE && bonustype != DMU_OT_SA)
+		return (ENOENT);
+	(void) fprintf(stderr, "modifying object that needs user accounting");
+	abort();
+	/* NOTREACHED */
+}
+
+/*
+ * Target is the dataset whose pool we want to open.
+ */
+static void
+import_pool(const char *target, boolean_t readonly)
+{
+	nvlist_t *config;
+	nvlist_t *pools;
+	int error;
+	char *sepp;
+	spa_t *spa;
+	nvpair_t *elem;
+	nvlist_t *props;
+	const char *name;
+
+	kernel_init(readonly ? FREAD : (FREAD | FWRITE));
+	g_zfs = libzfs_init();
+	ASSERT(g_zfs != NULL);
+
+	dmu_objset_register_type(DMU_OST_ZFS, space_delta_cb);
+
+	g_readonly = readonly;
+
+	/*
+	 * If we only want readonly access, it's OK if we find
+	 * a potentially-active (ie, imported into the kernel) pool from the
+	 * default cachefile.
+	 */
+	if (readonly && spa_open(target, &spa, FTAG) == 0) {
+		spa_close(spa, FTAG);
+		return;
+	}
+
+	g_importargs.unique = B_TRUE;
+	g_importargs.can_be_active = readonly;
+	g_pool = strdup(target);
+	if ((sepp = strpbrk(g_pool, "/@")) != NULL)
+		*sepp = '\0';
+	g_importargs.poolname = g_pool;
+	pools = zpool_search_import(g_zfs, &g_importargs);
+
+	if (pools == NULL || nvlist_next_nvpair(pools, NULL) == NULL) {
+		if (!g_importargs.can_be_active) {
+			g_importargs.can_be_active = B_TRUE;
+			if (zpool_search_import(g_zfs, &g_importargs) != NULL ||
+			    spa_open(target, &spa, FTAG) == 0) {
+				fatal("cannot import '%s': pool is active; run "
+				    "\"zpool export %s\" first\n",
+				    g_pool, g_pool);
+			}
+		}
+
+		fatal("cannot import '%s': no such pool available\n", g_pool);
+	}
+
+	elem = nvlist_next_nvpair(pools, NULL);
+	name = nvpair_name(elem);
+	verify(nvpair_value_nvlist(elem, &config) == 0);
+
+	props = NULL;
+	if (readonly) {
+		verify(nvlist_alloc(&props, NV_UNIQUE_NAME, 0) == 0);
+		verify(nvlist_add_uint64(props,
+		    zpool_prop_to_name(ZPOOL_PROP_READONLY), 1) == 0);
+	}
+
+	zfeature_checks_disable = B_TRUE;
+	error = spa_import(name, config, props, ZFS_IMPORT_NORMAL);
+	zfeature_checks_disable = B_FALSE;
+	if (error == EEXIST)
+		error = 0;
+
+	if (error)
+		fatal("can't import '%s': %s", name, strerror(error));
+}
+
+static void
+zhack_spa_open(const char *target, boolean_t readonly, void *tag, spa_t **spa)
+{
+	int err;
+
+	import_pool(target, readonly);
+
+	zfeature_checks_disable = B_TRUE;
+	err = spa_open(target, spa, tag);
+	zfeature_checks_disable = B_FALSE;
+
+	if (err != 0)
+		fatal("cannot open '%s': %s", target, strerror(err));
+	if (spa_version(*spa) < SPA_VERSION_FEATURES) {
+		fatal("'%s' has version %d, features not enabled", target,
+		    (int)spa_version(*spa));
+	}
+}
+
+static void
+dump_obj(objset_t *os, uint64_t obj, const char *name)
+{
+	zap_cursor_t zc;
+	zap_attribute_t za;
+
+	(void) printf("%s_obj:\n", name);
+
+	for (zap_cursor_init(&zc, os, obj);
+	    zap_cursor_retrieve(&zc, &za) == 0;
+	    zap_cursor_advance(&zc)) {
+		if (za.za_integer_length == 8) {
+			ASSERT(za.za_num_integers == 1);
+			(void) printf("\t%s = %llu\n",
+			    za.za_name, (u_longlong_t)za.za_first_integer);
+		} else {
+			ASSERT(za.za_integer_length == 1);
+			char val[1024];
+			VERIFY(zap_lookup(os, obj, za.za_name,
+			    1, sizeof (val), val) == 0);
+			(void) printf("\t%s = %s\n", za.za_name, val);
+		}
+	}
+	zap_cursor_fini(&zc);
+}
+
+static void
+dump_mos(spa_t *spa)
+{
+	nvlist_t *nv = spa->spa_label_features;
+
+	(void) printf("label config:\n");
+	for (nvpair_t *pair = nvlist_next_nvpair(nv, NULL);
+	    pair != NULL;
+	    pair = nvlist_next_nvpair(nv, pair)) {
+		(void) printf("\t%s\n", nvpair_name(pair));
+	}
+}
+
+static void
+zhack_do_feature_stat(int argc, char **argv)
+{
+	spa_t *spa;
+	objset_t *os;
+	char *target;
+
+	argc--;
+	argv++;
+
+	if (argc < 1) {
+		(void) fprintf(stderr, "error: missing pool name\n");
+		usage();
+	}
+	target = argv[0];
+
+	zhack_spa_open(target, B_TRUE, FTAG, &spa);
+	os = spa->spa_meta_objset;
+
+	dump_obj(os, spa->spa_feat_for_read_obj, "for_read");
+	dump_obj(os, spa->spa_feat_for_write_obj, "for_write");
+	dump_obj(os, spa->spa_feat_desc_obj, "descriptions");
+	dump_mos(spa);
+
+	spa_close(spa, FTAG);
+}
+
+static void
+feature_enable_sync(void *arg1, void *arg2, dmu_tx_t *tx)
+{
+	spa_t *spa = arg1;
+	zfeature_info_t *feature = arg2;
+
+	spa_feature_enable(spa, feature, tx);
+}
+
+static void
+zhack_do_feature_enable(int argc, char **argv)
+{
+	char c;
+	char *desc, *target;
+	spa_t *spa;
+	objset_t *mos;
+	zfeature_info_t feature;
+	zfeature_info_t *nodeps[] = { NULL };
+
+	/*
+	 * Features are not added to the pool's label until their refcounts
+	 * are incremented, so fi_mos can just be left as false for now.
+	 */
+	desc = NULL;
+	feature.fi_uname = "zhack";
+	feature.fi_mos = B_FALSE;
+	feature.fi_can_readonly = B_FALSE;
+	feature.fi_depends = nodeps;
+
+	optind = 1;
+	while ((c = getopt(argc, argv, "rmd:")) != -1) {
+		switch (c) {
+		case 'r':
+			feature.fi_can_readonly = B_TRUE;
+			break;
+		case 'd':
+			desc = strdup(optarg);
+			break;
+		default:
+			usage();
+			break;
+		}
+	}
+
+	if (desc == NULL)
+		desc = strdup("zhack injected");
+	feature.fi_desc = desc;
+
+	argc -= optind;
+	argv += optind;
+
+	if (argc < 2) {
+		(void) fprintf(stderr, "error: missing feature or pool name\n");
+		usage();
+	}
+	target = argv[0];
+	feature.fi_guid = argv[1];
+
+	if (!zfeature_is_valid_guid(feature.fi_guid))
+		fatal("invalid feature guid: %s", feature.fi_guid);
+
+	zhack_spa_open(target, B_FALSE, FTAG, &spa);
+	mos = spa->spa_meta_objset;
+
+	if (0 == zfeature_lookup_guid(feature.fi_guid, NULL))
+		fatal("'%s' is a real feature, will not enable");
+	if (0 == zap_contains(mos, spa->spa_feat_desc_obj, feature.fi_guid))
+		fatal("feature already enabled: %s", feature.fi_guid);
+
+	VERIFY3U(0, ==, dsl_sync_task_do(spa->spa_dsl_pool, NULL,
+	    feature_enable_sync, spa, &feature, 5));
+
+	spa_close(spa, FTAG);
+
+	free(desc);
+}
+
+static void
+feature_incr_sync(void *arg1, void *arg2, dmu_tx_t *tx)
+{
+	spa_t *spa = arg1;
+	zfeature_info_t *feature = arg2;
+
+	spa_feature_incr(spa, feature, tx);
+}
+
+static void
+feature_decr_sync(void *arg1, void *arg2, dmu_tx_t *tx)
+{
+	spa_t *spa = arg1;
+	zfeature_info_t *feature = arg2;
+
+	spa_feature_decr(spa, feature, tx);
+}
+
+static void
+zhack_do_feature_ref(int argc, char **argv)
+{
+	char c;
+	char *target;
+	boolean_t decr = B_FALSE;
+	spa_t *spa;
+	objset_t *mos;
+	zfeature_info_t feature;
+	zfeature_info_t *nodeps[] = { NULL };
+
+	/*
+	 * fi_desc does not matter here because it was written to disk
+	 * when the feature was enabled, but we need to properly set the
+	 * feature for read or write based on the information we read off
+	 * disk later.
+	 */
+	feature.fi_uname = "zhack";
+	feature.fi_mos = B_FALSE;
+	feature.fi_desc = NULL;
+	feature.fi_depends = nodeps;
+
+	optind = 1;
+	while ((c = getopt(argc, argv, "md")) != -1) {
+		switch (c) {
+		case 'm':
+			feature.fi_mos = B_TRUE;
+			break;
+		case 'd':
+			decr = B_TRUE;
+			break;
+		default:
+			usage();
+			break;
+		}
+	}
+	argc -= optind;
+	argv += optind;
+
+	if (argc < 2) {
+		(void) fprintf(stderr, "error: missing feature or pool name\n");
+		usage();
+	}
+	target = argv[0];
+	feature.fi_guid = argv[1];
+
+	if (!zfeature_is_valid_guid(feature.fi_guid))
+		fatal("invalid feature guid: %s", feature.fi_guid);
+
+	zhack_spa_open(target, B_FALSE, FTAG, &spa);
+	mos = spa->spa_meta_objset;
+
+	if (0 == zfeature_lookup_guid(feature.fi_guid, NULL))
+		fatal("'%s' is a real feature, will not change refcount");
+
+	if (0 == zap_contains(mos, spa->spa_feat_for_read_obj,
+	    feature.fi_guid)) {
+		feature.fi_can_readonly = B_FALSE;
+	} else if (0 == zap_contains(mos, spa->spa_feat_for_write_obj,
+	    feature.fi_guid)) {
+		feature.fi_can_readonly = B_TRUE;
+	} else {
+		fatal("feature is not enabled: %s", feature.fi_guid);
+	}
+
+	if (decr && !spa_feature_is_active(spa, &feature))
+		fatal("feature refcount already 0: %s", feature.fi_guid);
+
+	VERIFY3U(0, ==, dsl_sync_task_do(spa->spa_dsl_pool, NULL,
+	    decr ? feature_decr_sync : feature_incr_sync, spa, &feature, 5));
+
+	spa_close(spa, FTAG);
+}
+
+static int
+zhack_do_feature(int argc, char **argv)
+{
+	char *subcommand;
+
+	argc--;
+	argv++;
+	if (argc == 0) {
+		(void) fprintf(stderr,
+		    "error: no feature operation specified\n");
+		usage();
+	}
+
+	subcommand = argv[0];
+	if (strcmp(subcommand, "stat") == 0) {
+		zhack_do_feature_stat(argc, argv);
+	} else if (strcmp(subcommand, "enable") == 0) {
+		zhack_do_feature_enable(argc, argv);
+	} else if (strcmp(subcommand, "ref") == 0) {
+		zhack_do_feature_ref(argc, argv);
+	} else {
+		(void) fprintf(stderr, "error: unknown subcommand: %s\n",
+		    subcommand);
+		usage();
+	}
+
+	return (0);
+}
+
+#define	MAX_NUM_PATHS 1024
+
+int
+main(int argc, char **argv)
+{
+	extern void zfs_prop_init(void);
+
+	char *path[MAX_NUM_PATHS];
+	const char *subcommand;
+	int rv = 0;
+	char c;
+
+	g_importargs.path = path;
+
+	dprintf_setup(&argc, argv);
+	zfs_prop_init();
+
+	while ((c = getopt(argc, argv, "c:d:")) != -1) {
+		switch (c) {
+		case 'c':
+			g_importargs.cachefile = optarg;
+			break;
+		case 'd':
+			assert(g_importargs.paths < MAX_NUM_PATHS);
+			g_importargs.path[g_importargs.paths++] = optarg;
+			break;
+		default:
+			usage();
+			break;
+		}
+	}
+
+	argc -= optind;
+	argv += optind;
+	optind = 1;
+
+	if (argc == 0) {
+		(void) fprintf(stderr, "error: no command specified\n");
+		usage();
+	}
+
+	subcommand = argv[0];
+
+	if (strcmp(subcommand, "feature") == 0) {
+		rv = zhack_do_feature(argc, argv);
+	} else {
+		(void) fprintf(stderr, "error: unknown subcommand: %s\n",
+		    subcommand);
+		usage();
+	}
+
+	if (!g_readonly && spa_export(g_pool, NULL, B_TRUE, B_TRUE) != 0) {
+		fatal("pool export failed; "
+		    "changes may not be committed to disk\n");
+	}
+
+	libzfs_fini(g_zfs);
+	kernel_fini();
+
+	return (rv);
+}

cmd/zinject/translate.c

@@ -0,0 +1,494 @@
+/*
+ * 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 (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#include <libzfs.h>
+
+#include <sys/zfs_context.h>
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdarg.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+#include <sys/file.h>
+#include <sys/mntent.h>
+#include <sys/mnttab.h>
+#include <sys/param.h>
+#include <sys/stat.h>
+
+#include <sys/dmu.h>
+#include <sys/dmu_objset.h>
+#include <sys/dnode.h>
+#include <sys/vdev_impl.h>
+
+#include <sys/mkdev.h>
+
+#include "zinject.h"
+
+extern void kernel_init(int);
+extern void kernel_fini(void);
+
+static int debug;
+
+static void
+ziprintf(const char *fmt, ...)
+{
+	va_list ap;
+
+	if (!debug)
+		return;
+
+	va_start(ap, fmt);
+	(void) vprintf(fmt, ap);
+	va_end(ap);
+}
+
+static void
+compress_slashes(const char *src, char *dest)
+{
+	while (*src != '\0') {
+		*dest = *src++;
+		while (*dest == '/' && *src == '/')
+			++src;
+		++dest;
+	}
+	*dest = '\0';
+}
+
+/*
+ * Given a full path to a file, translate into a dataset name and a relative
+ * path within the dataset.  'dataset' must be at least MAXNAMELEN characters,
+ * and 'relpath' must be at least MAXPATHLEN characters.  We also pass a stat64
+ * buffer, which we need later to get the object ID.
+ */
+static int
+parse_pathname(const char *inpath, char *dataset, char *relpath,
+    struct stat64 *statbuf)
+{
+	struct extmnttab mp;
+	FILE *fp;
+	int match;
+	const char *rel;
+	char fullpath[MAXPATHLEN];
+
+	compress_slashes(inpath, fullpath);
+
+	if (fullpath[0] != '/') {
+		(void) fprintf(stderr, "invalid object '%s': must be full "
+		    "path\n", fullpath);
+		usage();
+		return (-1);
+	}
+
+	if (strlen(fullpath) >= MAXPATHLEN) {
+		(void) fprintf(stderr, "invalid object; pathname too long\n");
+		return (-1);
+	}
+
+	if (stat64(fullpath, statbuf) != 0) {
+		(void) fprintf(stderr, "cannot open '%s': %s\n",
+		    fullpath, strerror(errno));
+		return (-1);
+	}
+
+	if ((fp = fopen(MNTTAB, "r")) == NULL) {
+		(void) fprintf(stderr, "cannot open /etc/mnttab\n");
+		return (-1);
+	}
+
+	match = 0;
+	while (getextmntent(fp, &mp, sizeof (mp)) == 0) {
+		if (makedev(mp.mnt_major, mp.mnt_minor) == statbuf->st_dev) {
+			match = 1;
+			break;
+		}
+	}
+
+	if (!match) {
+		(void) fprintf(stderr, "cannot find mountpoint for '%s'\n",
+		    fullpath);
+		return (-1);
+	}
+
+	if (strcmp(mp.mnt_fstype, MNTTYPE_ZFS) != 0) {
+		(void) fprintf(stderr, "invalid path '%s': not a ZFS "
+		    "filesystem\n", fullpath);
+		return (-1);
+	}
+
+	if (strncmp(fullpath, mp.mnt_mountp, strlen(mp.mnt_mountp)) != 0) {
+		(void) fprintf(stderr, "invalid path '%s': mountpoint "
+		    "doesn't match path\n", fullpath);
+		return (-1);
+	}
+
+	(void) strcpy(dataset, mp.mnt_special);
+
+	rel = fullpath + strlen(mp.mnt_mountp);
+	if (rel[0] == '/')
+		rel++;
+	(void) strcpy(relpath, rel);
+
+	return (0);
+}
+
+/*
+ * Convert from a (dataset, path) pair into a (objset, object) pair.  Note that
+ * we grab the object number from the inode number, since looking this up via
+ * libzpool is a real pain.
+ */
+/* ARGSUSED */
+static int
+object_from_path(const char *dataset, const char *path, struct stat64 *statbuf,
+    zinject_record_t *record)
+{
+	objset_t *os;
+	int err;
+
+	/*
+	 * Before doing any libzpool operations, call sync() to ensure that the
+	 * on-disk state is consistent with the in-core state.
+	 */
+	sync();
+
+	err = dmu_objset_own(dataset, DMU_OST_ZFS, B_TRUE, FTAG, &os);
+	if (err != 0) {
+		(void) fprintf(stderr, "cannot open dataset '%s': %s\n",
+		    dataset, strerror(err));
+		return (-1);
+	}
+
+	record->zi_objset = dmu_objset_id(os);
+	record->zi_object = statbuf->st_ino;
+
+	dmu_objset_disown(os, FTAG);
+
+	return (0);
+}
+
+/*
+ * Calculate the real range based on the type, level, and range given.
+ */
+static int
+calculate_range(const char *dataset, err_type_t type, int level, char *range,
+    zinject_record_t *record)
+{
+	objset_t *os = NULL;
+	dnode_t *dn = NULL;
+	int err;
+	int ret = -1;
+
+	/*
+	 * Determine the numeric range from the string.
+	 */
+	if (range == NULL) {
+		/*
+		 * If range is unspecified, set the range to [0,-1], which
+		 * indicates that the whole object should be treated as an
+		 * error.
+		 */
+		record->zi_start = 0;
+		record->zi_end = -1ULL;
+	} else {
+		char *end;
+
+		/* XXX add support for suffixes */
+		record->zi_start = strtoull(range, &end, 10);
+
+
+		if (*end == '\0')
+			record->zi_end = record->zi_start + 1;
+		else if (*end == ',')
+			record->zi_end = strtoull(end + 1, &end, 10);
+
+		if (*end != '\0') {
+			(void) fprintf(stderr, "invalid range '%s': must be "
+			    "a numeric range of the form 'start[,end]'\n",
+			    range);
+			goto out;
+		}
+	}
+
+	switch (type) {
+	case TYPE_DATA:
+		break;
+
+	case TYPE_DNODE:
+		/*
+		 * If this is a request to inject faults into the dnode, then we
+		 * must translate the current (objset,object) pair into an
+		 * offset within the metadnode for the objset.  Specifying any
+		 * kind of range with type 'dnode' is illegal.
+		 */
+		if (range != NULL) {
+			(void) fprintf(stderr, "range cannot be specified when "
+			    "type is 'dnode'\n");
+			goto out;
+		}
+
+		record->zi_start = record->zi_object * sizeof (dnode_phys_t);
+		record->zi_end = record->zi_start + sizeof (dnode_phys_t);
+		record->zi_object = 0;
+		break;
+	}
+
+	/*
+	 * Get the dnode associated with object, so we can calculate the block
+	 * size.
+	 */
+	if ((err = dmu_objset_own(dataset, DMU_OST_ANY,
+	    B_TRUE, FTAG, &os)) != 0) {
+		(void) fprintf(stderr, "cannot open dataset '%s': %s\n",
+		    dataset, strerror(err));
+		goto out;
+	}
+
+	if (record->zi_object == 0) {
+		dn = DMU_META_DNODE(os);
+	} else {
+		err = dnode_hold(os, record->zi_object, FTAG, &dn);
+		if (err != 0) {
+			(void) fprintf(stderr, "failed to hold dnode "
+			    "for object %llu\n",
+			    (u_longlong_t)record->zi_object);
+			goto out;
+		}
+	}
+
+
+	ziprintf("data shift: %d\n", (int)dn->dn_datablkshift);
+	ziprintf(" ind shift: %d\n", (int)dn->dn_indblkshift);
+
+	/*
+	 * Translate range into block IDs.
+	 */
+	if (record->zi_start != 0 || record->zi_end != -1ULL) {
+		record->zi_start >>= dn->dn_datablkshift;
+		record->zi_end >>= dn->dn_datablkshift;
+	}
+
+	/*
+	 * Check level, and then translate level 0 blkids into ranges
+	 * appropriate for level of indirection.
+	 */
+	record->zi_level = level;
+	if (level > 0) {
+		ziprintf("level 0 blkid range: [%llu, %llu]\n",
+		    record->zi_start, record->zi_end);
+
+		if (level >= dn->dn_nlevels) {
+			(void) fprintf(stderr, "level %d exceeds max level "
+			    "of object (%d)\n", level, dn->dn_nlevels - 1);
+			goto out;
+		}
+
+		if (record->zi_start != 0 || record->zi_end != 0) {
+			int shift = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
+
+			for (; level > 0; level--) {
+				record->zi_start >>= shift;
+				record->zi_end >>= shift;
+			}
+		}
+	}
+
+	ret = 0;
+out:
+	if (dn) {
+		if (dn != DMU_META_DNODE(os))
+			dnode_rele(dn, FTAG);
+	}
+	if (os)
+		dmu_objset_disown(os, FTAG);
+
+	return (ret);
+}
+
+int
+translate_record(err_type_t type, const char *object, const char *range,
+    int level, zinject_record_t *record, char *poolname, char *dataset)
+{
+	char path[MAXPATHLEN];
+	char *slash;
+	struct stat64 statbuf;
+	int ret = -1;
+
+	kernel_init(FREAD);
+
+	debug = (getenv("ZINJECT_DEBUG") != NULL);
+
+	ziprintf("translating: %s\n", object);
+
+	if (MOS_TYPE(type)) {
+		/*
+		 * MOS objects are treated specially.
+		 */
+		switch (type) {
+		case TYPE_MOS:
+			record->zi_type = 0;
+			break;
+		case TYPE_MOSDIR:
+			record->zi_type = DMU_OT_OBJECT_DIRECTORY;
+			break;
+		case TYPE_METASLAB:
+			record->zi_type = DMU_OT_OBJECT_ARRAY;
+			break;
+		case TYPE_CONFIG:
+			record->zi_type = DMU_OT_PACKED_NVLIST;
+			break;
+		case TYPE_BPOBJ:
+			record->zi_type = DMU_OT_BPOBJ;
+			break;
+		case TYPE_SPACEMAP:
+			record->zi_type = DMU_OT_SPACE_MAP;
+			break;
+		case TYPE_ERRLOG:
+			record->zi_type = DMU_OT_ERROR_LOG;
+			break;
+		}
+
+		dataset[0] = '\0';
+		(void) strcpy(poolname, object);
+		return (0);
+	}
+
+	/*
+	 * Convert a full path into a (dataset, file) pair.
+	 */
+	if (parse_pathname(object, dataset, path, &statbuf) != 0)
+		goto err;
+
+	ziprintf("   dataset: %s\n", dataset);
+	ziprintf("      path: %s\n", path);
+
+	/*
+	 * Convert (dataset, file) into (objset, object)
+	 */
+	if (object_from_path(dataset, path, &statbuf, record) != 0)
+		goto err;
+
+	ziprintf("raw objset: %llu\n", record->zi_objset);
+	ziprintf("raw object: %llu\n", record->zi_object);
+
+	/*
+	 * For the given object, calculate the real (type, level, range)
+	 */
+	if (calculate_range(dataset, type, level, (char *)range, record) != 0)
+		goto err;
+
+	ziprintf("    objset: %llu\n", record->zi_objset);
+	ziprintf("    object: %llu\n", record->zi_object);
+	if (record->zi_start == 0 &&
+	    record->zi_end == -1ULL)
+		ziprintf("     range: all\n");
+	else
+		ziprintf("     range: [%llu, %llu]\n", record->zi_start,
+		    record->zi_end);
+
+	/*
+	 * Copy the pool name
+	 */
+	(void) strcpy(poolname, dataset);
+	if ((slash = strchr(poolname, '/')) != NULL)
+		*slash = '\0';
+
+	ret = 0;
+
+err:
+	kernel_fini();
+	return (ret);
+}
+
+int
+translate_raw(const char *str, zinject_record_t *record)
+{
+	/*
+	 * A raw bookmark of the form objset:object:level:blkid, where each
+	 * number is a hexidecimal value.
+	 */
+	if (sscanf(str, "%llx:%llx:%x:%llx", (u_longlong_t *)&record->zi_objset,
+	    (u_longlong_t *)&record->zi_object, &record->zi_level,
+	    (u_longlong_t *)&record->zi_start) != 4) {
+		(void) fprintf(stderr, "bad raw spec '%s': must be of the form "
+		    "'objset:object:level:blkid'\n", str);
+		return (-1);
+	}
+
+	record->zi_end = record->zi_start;
+
+	return (0);
+}
+
+int
+translate_device(const char *pool, const char *device, err_type_t label_type,
+    zinject_record_t *record)
+{
+	char *end;
+	zpool_handle_t *zhp;
+	nvlist_t *tgt;
+	boolean_t isspare, iscache;
+
+	/*
+	 * Given a device name or GUID, create an appropriate injection record
+	 * with zi_guid set.
+	 */
+	if ((zhp = zpool_open(g_zfs, pool)) == NULL)
+		return (-1);
+
+	record->zi_guid = strtoull(device, &end, 16);
+	if (record->zi_guid == 0 || *end != '\0') {
+		tgt = zpool_find_vdev(zhp, device, &isspare, &iscache, NULL);
+
+		if (tgt == NULL) {
+			(void) fprintf(stderr, "cannot find device '%s' in "
+			    "pool '%s'\n", device, pool);
+			return (-1);
+		}
+
+		verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID,
+		    &record->zi_guid) == 0);
+	}
+
+	switch (label_type) {
+	case TYPE_LABEL_UBERBLOCK:
+		record->zi_start = offsetof(vdev_label_t, vl_uberblock[0]);
+		record->zi_end = record->zi_start + VDEV_UBERBLOCK_RING - 1;
+		break;
+	case TYPE_LABEL_NVLIST:
+		record->zi_start = offsetof(vdev_label_t, vl_vdev_phys);
+		record->zi_end = record->zi_start + VDEV_PHYS_SIZE - 1;
+		break;
+	case TYPE_LABEL_PAD1:
+		record->zi_start = offsetof(vdev_label_t, vl_pad1);
+		record->zi_end = record->zi_start + VDEV_PAD_SIZE - 1;
+		break;
+	case TYPE_LABEL_PAD2:
+		record->zi_start = offsetof(vdev_label_t, vl_pad2);
+		record->zi_end = record->zi_start + VDEV_PAD_SIZE - 1;
+		break;
+	}
+	return (0);
+}

cmd/zinject/zinject.c

@@ -0,0 +1,972 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+/*
+ * ZFS Fault Injector
+ *
+ * This userland component takes a set of options and uses libzpool to translate
+ * from a user-visible object type and name to an internal representation.
+ * There are two basic types of faults: device faults and data faults.
+ *
+ *
+ * DEVICE FAULTS
+ *
+ * Errors can be injected into a particular vdev using the '-d' option.  This
+ * option takes a path or vdev GUID to uniquely identify the device within a
+ * pool.  There are two types of errors that can be injected, EIO and ENXIO,
+ * that can be controlled through the '-e' option.  The default is ENXIO.  For
+ * EIO failures, any attempt to read data from the device will return EIO, but
+ * subsequent attempt to reopen the device will succeed.  For ENXIO failures,
+ * any attempt to read from the device will return EIO, but any attempt to
+ * reopen the device will also return ENXIO.
+ * For label faults, the -L option must be specified. This allows faults
+ * to be injected into either the nvlist, uberblock, pad1, or pad2 region
+ * of all the labels for the specified device.
+ *
+ * This form of the command looks like:
+ *
+ * 	zinject -d device [-e errno] [-L <uber | nvlist | pad1 | pad2>] pool
+ *
+ *
+ * DATA FAULTS
+ *
+ * We begin with a tuple of the form:
+ *
+ * 	<type,level,range,object>
+ *
+ * 	type	A string describing the type of data to target.  Each type
+ * 		implicitly describes how to interpret 'object'. Currently,
+ * 		the following values are supported:
+ *
+ * 		data		User data for a file
+ * 		dnode		Dnode for a file or directory
+ *
+ *		The following MOS objects are special.  Instead of injecting
+ *		errors on a particular object or blkid, we inject errors across
+ *		all objects of the given type.
+ *
+ * 		mos		Any data in the MOS
+ * 		mosdir		object directory
+ * 		config		pool configuration
+ * 		bpobj		blkptr list
+ * 		spacemap	spacemap
+ * 		metaslab	metaslab
+ * 		errlog		persistent error log
+ *
+ * 	level	Object level.  Defaults to '0', not applicable to all types.  If
+ * 		a range is given, this corresponds to the indirect block
+ * 		corresponding to the specific range.
+ *
+ *	range	A numerical range [start,end) within the object.  Defaults to
+ *		the full size of the file.
+ *
+ * 	object	A string describing the logical location of the object.  For
+ * 		files and directories (currently the only supported types),
+ * 		this is the path of the object on disk.
+ *
+ * This is translated, via libzpool, into the following internal representation:
+ *
+ * 	<type,objset,object,level,range>
+ *
+ * These types should be self-explanatory.  This tuple is then passed to the
+ * kernel via a special ioctl() to initiate fault injection for the given
+ * object.  Note that 'type' is not strictly necessary for fault injection, but
+ * is used when translating existing faults into a human-readable string.
+ *
+ *
+ * The command itself takes one of the forms:
+ *
+ * 	zinject
+ * 	zinject <-a | -u pool>
+ * 	zinject -c <id|all>
+ * 	zinject [-q] <-t type> [-f freq] [-u] [-a] [-m] [-e errno] [-l level]
+ *	    [-r range] <object>
+ * 	zinject [-f freq] [-a] [-m] [-u] -b objset:object:level:start:end pool
+ *
+ * With no arguments, the command prints all currently registered injection
+ * handlers, with their numeric identifiers.
+ *
+ * The '-c' option will clear the given handler, or all handlers if 'all' is
+ * specified.
+ *
+ * The '-e' option takes a string describing the errno to simulate.  This must
+ * be either 'io' or 'checksum'.  In most cases this will result in the same
+ * behavior, but RAID-Z will produce a different set of ereports for this
+ * situation.
+ *
+ * The '-a', '-u', and '-m' flags toggle internal flush behavior.  If '-a' is
+ * specified, then the ARC cache is flushed appropriately.  If '-u' is
+ * specified, then the underlying SPA is unloaded.  Either of these flags can be
+ * specified independently of any other handlers.  The '-m' flag automatically
+ * does an unmount and remount of the underlying dataset to aid in flushing the
+ * cache.
+ *
+ * The '-f' flag controls the frequency of errors injected, expressed as a
+ * integer percentage between 1 and 100.  The default is 100.
+ *
+ * The this form is responsible for actually injecting the handler into the
+ * framework.  It takes the arguments described above, translates them to the
+ * internal tuple using libzpool, and then issues an ioctl() to register the
+ * handler.
+ *
+ * The final form can target a specific bookmark, regardless of whether a
+ * human-readable interface has been designed.  It allows developers to specify
+ * a particular block by number.
+ */
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+#include <unistd.h>
+
+#include <sys/fs/zfs.h>
+#include <sys/mount.h>
+
+#include <libzfs.h>
+
+#undef verify	/* both libzfs.h and zfs_context.h want to define this */
+
+#include "zinject.h"
+
+libzfs_handle_t *g_zfs;
+int zfs_fd;
+
+#define	ECKSUM	EBADE
+
+static const char *errtable[TYPE_INVAL] = {
+	"data",
+	"dnode",
+	"mos",
+	"mosdir",
+	"metaslab",
+	"config",
+	"bpobj",
+	"spacemap",
+	"errlog",
+	"uber",
+	"nvlist",
+	"pad1",
+	"pad2"
+};
+
+static err_type_t
+name_to_type(const char *arg)
+{
+	int i;
+	for (i = 0; i < TYPE_INVAL; i++)
+		if (strcmp(errtable[i], arg) == 0)
+			return (i);
+
+	return (TYPE_INVAL);
+}
+
+static const char *
+type_to_name(uint64_t type)
+{
+	switch (type) {
+	case DMU_OT_OBJECT_DIRECTORY:
+		return ("mosdir");
+	case DMU_OT_OBJECT_ARRAY:
+		return ("metaslab");
+	case DMU_OT_PACKED_NVLIST:
+		return ("config");
+	case DMU_OT_BPOBJ:
+		return ("bpobj");
+	case DMU_OT_SPACE_MAP:
+		return ("spacemap");
+	case DMU_OT_ERROR_LOG:
+		return ("errlog");
+	default:
+		return ("-");
+	}
+}
+
+
+/*
+ * Print usage message.
+ */
+void
+usage(void)
+{
+	(void) printf(
+	    "usage:\n"
+	    "\n"
+	    "\tzinject\n"
+	    "\n"
+	    "\t\tList all active injection records.\n"
+	    "\n"
+	    "\tzinject -c <id|all>\n"
+	    "\n"
+	    "\t\tClear the particular record (if given a numeric ID), or\n"
+	    "\t\tall records if 'all' is specificed.\n"
+	    "\n"
+	    "\tzinject -p <function name> pool\n"
+	    "\t\tInject a panic fault at the specified function. Only \n"
+	    "\t\tfunctions which call spa_vdev_config_exit(), or \n"
+	    "\t\tspa_vdev_exit() will trigger a panic.\n"
+	    "\n"
+	    "\tzinject -d device [-e errno] [-L <nvlist|uber|pad1|pad2>] [-F]\n"
+	    "\t    [-T <read|write|free|claim|all> pool\n"
+	    "\t\tInject a fault into a particular device or the device's\n"
+	    "\t\tlabel.  Label injection can either be 'nvlist', 'uber',\n "
+	    "\t\t'pad1', or 'pad2'.\n"
+	    "\t\t'errno' can be 'nxio' (the default), 'io', or 'dtl'.\n"
+	    "\n"
+	    "\tzinject -d device -A <degrade|fault> pool\n"
+	    "\t\tPerform a specific action on a particular device\n"
+	    "\n"
+	    "\tzinject -I [-s <seconds> | -g <txgs>] pool\n"
+	    "\t\tCause the pool to stop writing blocks yet not\n"
+	    "\t\treport errors for a duration.  Simulates buggy hardware\n"
+	    "\t\tthat fails to honor cache flush requests.\n"
+	    "\t\tDefault duration is 30 seconds.  The machine is panicked\n"
+	    "\t\tat the end of the duration.\n"
+	    "\n"
+	    "\tzinject -b objset:object:level:blkid pool\n"
+	    "\n"
+	    "\t\tInject an error into pool 'pool' with the numeric bookmark\n"
+	    "\t\tspecified by the remaining tuple.  Each number is in\n"
+	    "\t\thexidecimal, and only one block can be specified.\n"
+	    "\n"
+	    "\tzinject [-q] <-t type> [-e errno] [-l level] [-r range]\n"
+	    "\t    [-a] [-m] [-u] [-f freq] <object>\n"
+	    "\n"
+	    "\t\tInject an error into the object specified by the '-t' option\n"
+	    "\t\tand the object descriptor.  The 'object' parameter is\n"
+	    "\t\tinterperted depending on the '-t' option.\n"
+	    "\n"
+	    "\t\t-q\tQuiet mode.  Only print out the handler number added.\n"
+	    "\t\t-e\tInject a specific error.  Must be either 'io' or\n"
+	    "\t\t\t'checksum'.  Default is 'io'.\n"
+	    "\t\t-l\tInject error at a particular block level. Default is "
+	    "0.\n"
+	    "\t\t-m\tAutomatically remount underlying filesystem.\n"
+	    "\t\t-r\tInject error over a particular logical range of an\n"
+	    "\t\t\tobject.  Will be translated to the appropriate blkid\n"
+	    "\t\t\trange according to the object's properties.\n"
+	    "\t\t-a\tFlush the ARC cache.  Can be specified without any\n"
+	    "\t\t\tassociated object.\n"
+	    "\t\t-u\tUnload the associated pool.  Can be specified with only\n"
+	    "\t\t\ta pool object.\n"
+	    "\t\t-f\tOnly inject errors a fraction of the time.  Expressed as\n"
+	    "\t\t\ta percentage between 1 and 100.\n"
+	    "\n"
+	    "\t-t data\t\tInject an error into the plain file contents of a\n"
+	    "\t\t\tfile.  The object must be specified as a complete path\n"
+	    "\t\t\tto a file on a ZFS filesystem.\n"
+	    "\n"
+	    "\t-t dnode\tInject an error into the metadnode in the block\n"
+	    "\t\t\tcorresponding to the dnode for a file or directory.  The\n"
+	    "\t\t\t'-r' option is incompatible with this mode.  The object\n"
+	    "\t\t\tis specified as a complete path to a file or directory\n"
+	    "\t\t\ton a ZFS filesystem.\n"
+	    "\n"
+	    "\t-t <mos>\tInject errors into the MOS for objects of the given\n"
+	    "\t\t\ttype.  Valid types are: mos, mosdir, config, bpobj,\n"
+	    "\t\t\tspacemap, metaslab, errlog.  The only valid <object> is\n"
+	    "\t\t\tthe poolname.\n");
+}
+
+static int
+iter_handlers(int (*func)(int, const char *, zinject_record_t *, void *),
+    void *data)
+{
+	zfs_cmd_t zc;
+	int ret;
+
+	zc.zc_guid = 0;
+
+	while (ioctl(zfs_fd, ZFS_IOC_INJECT_LIST_NEXT, &zc) == 0)
+		if ((ret = func((int)zc.zc_guid, zc.zc_name,
+		    &zc.zc_inject_record, data)) != 0)
+			return (ret);
+
+	if (errno != ENOENT) {
+		(void) fprintf(stderr, "Unable to list handlers: %s\n",
+		    strerror(errno));
+		return (-1);
+	}
+
+	return (0);
+}
+
+static int
+print_data_handler(int id, const char *pool, zinject_record_t *record,
+    void *data)
+{
+	int *count = data;
+
+	if (record->zi_guid != 0 || record->zi_func[0] != '\0')
+		return (0);
+
+	if (*count == 0) {
+		(void) printf("%3s  %-15s  %-6s  %-6s  %-8s  %3s  %-15s\n",
+		    "ID", "POOL", "OBJSET", "OBJECT", "TYPE", "LVL",  "RANGE");
+		(void) printf("---  ---------------  ------  "
+		    "------  --------  ---  ---------------\n");
+	}
+
+	*count += 1;
+
+	(void) printf("%3d  %-15s  %-6llu  %-6llu  %-8s  %3d  ", id, pool,
+	    (u_longlong_t)record->zi_objset, (u_longlong_t)record->zi_object,
+	    type_to_name(record->zi_type), record->zi_level);
+
+	if (record->zi_start == 0 &&
+	    record->zi_end == -1ULL)
+		(void) printf("all\n");
+	else
+		(void) printf("[%llu, %llu]\n", (u_longlong_t)record->zi_start,
+		    (u_longlong_t)record->zi_end);
+
+	return (0);
+}
+
+static int
+print_device_handler(int id, const char *pool, zinject_record_t *record,
+    void *data)
+{
+	int *count = data;
+
+	if (record->zi_guid == 0 || record->zi_func[0] != '\0')
+		return (0);
+
+	if (*count == 0) {
+		(void) printf("%3s  %-15s  %s\n", "ID", "POOL", "GUID");
+		(void) printf("---  ---------------  ----------------\n");
+	}
+
+	*count += 1;
+
+	(void) printf("%3d  %-15s  %llx\n", id, pool,
+	    (u_longlong_t)record->zi_guid);
+
+	return (0);
+}
+
+static int
+print_panic_handler(int id, const char *pool, zinject_record_t *record,
+    void *data)
+{
+	int *count = data;
+
+	if (record->zi_func[0] == '\0')
+		return (0);
+
+	if (*count == 0) {
+		(void) printf("%3s  %-15s  %s\n", "ID", "POOL", "FUNCTION");
+		(void) printf("---  ---------------  ----------------\n");
+	}
+
+	*count += 1;
+
+	(void) printf("%3d  %-15s  %s\n", id, pool, record->zi_func);
+
+	return (0);
+}
+
+/*
+ * Print all registered error handlers.  Returns the number of handlers
+ * registered.
+ */
+static int
+print_all_handlers(void)
+{
+	int count = 0, total = 0;
+
+	(void) iter_handlers(print_device_handler, &count);
+	if (count > 0) {
+		total += count;
+		(void) printf("\n");
+		count = 0;
+	}
+
+	(void) iter_handlers(print_data_handler, &count);
+	if (count > 0) {
+		total += count;
+		(void) printf("\n");
+		count = 0;
+	}
+
+	(void) iter_handlers(print_panic_handler, &count);
+
+	return (count + total);
+}
+
+/* ARGSUSED */
+static int
+cancel_one_handler(int id, const char *pool, zinject_record_t *record,
+    void *data)
+{
+	zfs_cmd_t zc;
+
+	zc.zc_guid = (uint64_t)id;
+
+	if (ioctl(zfs_fd, ZFS_IOC_CLEAR_FAULT, &zc) != 0) {
+		(void) fprintf(stderr, "failed to remove handler %d: %s\n",
+		    id, strerror(errno));
+		return (1);
+	}
+
+	return (0);
+}
+
+/*
+ * Remove all fault injection handlers.
+ */
+static int
+cancel_all_handlers(void)
+{
+	int ret = iter_handlers(cancel_one_handler, NULL);
+
+	if (ret == 0)
+		(void) printf("removed all registered handlers\n");
+
+	return (ret);
+}
+
+/*
+ * Remove a specific fault injection handler.
+ */
+static int
+cancel_handler(int id)
+{
+	zfs_cmd_t zc;
+
+	zc.zc_guid = (uint64_t)id;
+
+	if (ioctl(zfs_fd, ZFS_IOC_CLEAR_FAULT, &zc) != 0) {
+		(void) fprintf(stderr, "failed to remove handler %d: %s\n",
+		    id, strerror(errno));
+		return (1);
+	}
+
+	(void) printf("removed handler %d\n", id);
+
+	return (0);
+}
+
+/*
+ * Register a new fault injection handler.
+ */
+static int
+register_handler(const char *pool, int flags, zinject_record_t *record,
+    int quiet)
+{
+	zfs_cmd_t zc;
+
+	(void) strcpy(zc.zc_name, pool);
+	zc.zc_inject_record = *record;
+	zc.zc_guid = flags;
+
+	if (ioctl(zfs_fd, ZFS_IOC_INJECT_FAULT, &zc) != 0) {
+		(void) fprintf(stderr, "failed to add handler: %s\n",
+		    strerror(errno));
+		return (1);
+	}
+
+	if (flags & ZINJECT_NULL)
+		return (0);
+
+	if (quiet) {
+		(void) printf("%llu\n", (u_longlong_t)zc.zc_guid);
+	} else {
+		(void) printf("Added handler %llu with the following "
+		    "properties:\n", (u_longlong_t)zc.zc_guid);
+		(void) printf("  pool: %s\n", pool);
+		if (record->zi_guid) {
+			(void) printf("  vdev: %llx\n",
+			    (u_longlong_t)record->zi_guid);
+		} else if (record->zi_func[0] != '\0') {
+			(void) printf("  panic function: %s\n",
+			    record->zi_func);
+		} else if (record->zi_duration > 0) {
+			(void) printf(" time: %lld seconds\n",
+			    (u_longlong_t)record->zi_duration);
+		} else if (record->zi_duration < 0) {
+			(void) printf(" txgs: %lld \n",
+			    (u_longlong_t)-record->zi_duration);
+		} else {
+			(void) printf("objset: %llu\n",
+			    (u_longlong_t)record->zi_objset);
+			(void) printf("object: %llu\n",
+			    (u_longlong_t)record->zi_object);
+			(void) printf("  type: %llu\n",
+			    (u_longlong_t)record->zi_type);
+			(void) printf(" level: %d\n", record->zi_level);
+			if (record->zi_start == 0 &&
+			    record->zi_end == -1ULL)
+				(void) printf(" range: all\n");
+			else
+				(void) printf(" range: [%llu, %llu)\n",
+				    (u_longlong_t)record->zi_start,
+				    (u_longlong_t)record->zi_end);
+		}
+	}
+
+	return (0);
+}
+
+int
+perform_action(const char *pool, zinject_record_t *record, int cmd)
+{
+	zfs_cmd_t zc;
+
+	ASSERT(cmd == VDEV_STATE_DEGRADED || cmd == VDEV_STATE_FAULTED);
+	(void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name));
+	zc.zc_guid = record->zi_guid;
+	zc.zc_cookie = cmd;
+
+	if (ioctl(zfs_fd, ZFS_IOC_VDEV_SET_STATE, &zc) == 0)
+		return (0);
+
+	return (1);
+}
+
+int
+main(int argc, char **argv)
+{
+	int c;
+	char *range = NULL;
+	char *cancel = NULL;
+	char *end;
+	char *raw = NULL;
+	char *device = NULL;
+	int level = 0;
+	int quiet = 0;
+	int error = 0;
+	int domount = 0;
+	int io_type = ZIO_TYPES;
+	int action = VDEV_STATE_UNKNOWN;
+	err_type_t type = TYPE_INVAL;
+	err_type_t label = TYPE_INVAL;
+	zinject_record_t record = { 0 };
+	char pool[MAXNAMELEN];
+	char dataset[MAXNAMELEN];
+	zfs_handle_t *zhp;
+	int nowrites = 0;
+	int dur_txg = 0;
+	int dur_secs = 0;
+	int ret;
+	int flags = 0;
+
+	if ((g_zfs = libzfs_init()) == NULL) {
+		(void) fprintf(stderr, "internal error: failed to "
+		    "initialize ZFS library\n");
+		return (1);
+	}
+
+	libzfs_print_on_error(g_zfs, B_TRUE);
+
+	if ((zfs_fd = open(ZFS_DEV, O_RDWR)) < 0) {
+		(void) fprintf(stderr, "failed to open ZFS device\n");
+		return (1);
+	}
+
+	if (argc == 1) {
+		/*
+		 * No arguments.  Print the available handlers.  If there are no
+		 * available handlers, direct the user to '-h' for help
+		 * information.
+		 */
+		if (print_all_handlers() == 0) {
+			(void) printf("No handlers registered.\n");
+			(void) printf("Run 'zinject -h' for usage "
+			    "information.\n");
+		}
+
+		return (0);
+	}
+
+	while ((c = getopt(argc, argv,
+	    ":aA:b:d:f:Fg:qhIc:t:T:l:mr:s:e:uL:p:")) != -1) {
+		switch (c) {
+		case 'a':
+			flags |= ZINJECT_FLUSH_ARC;
+			break;
+		case 'A':
+			if (strcasecmp(optarg, "degrade") == 0) {
+				action = VDEV_STATE_DEGRADED;
+			} else if (strcasecmp(optarg, "fault") == 0) {
+				action = VDEV_STATE_FAULTED;
+			} else {
+				(void) fprintf(stderr, "invalid action '%s': "
+				    "must be 'degrade' or 'fault'\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 'b':
+			raw = optarg;
+			break;
+		case 'c':
+			cancel = optarg;
+			break;
+		case 'd':
+			device = optarg;
+			break;
+		case 'e':
+			if (strcasecmp(optarg, "io") == 0) {
+				error = EIO;
+			} else if (strcasecmp(optarg, "checksum") == 0) {
+				error = ECKSUM;
+			} else if (strcasecmp(optarg, "nxio") == 0) {
+				error = ENXIO;
+			} else if (strcasecmp(optarg, "dtl") == 0) {
+				error = ECHILD;
+			} else {
+				(void) fprintf(stderr, "invalid error type "
+				    "'%s': must be 'io', 'checksum' or "
+				    "'nxio'\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 'f':
+			record.zi_freq = atoi(optarg);
+			if (record.zi_freq < 1 || record.zi_freq > 100) {
+				(void) fprintf(stderr, "frequency range must "
+				    "be in the range (0, 100]\n");
+				return (1);
+			}
+			break;
+		case 'F':
+			record.zi_failfast = B_TRUE;
+			break;
+		case 'g':
+			dur_txg = 1;
+			record.zi_duration = (int)strtol(optarg, &end, 10);
+			if (record.zi_duration <= 0 || *end != '\0') {
+				(void) fprintf(stderr, "invalid duration '%s': "
+				    "must be a positive integer\n", optarg);
+				usage();
+				return (1);
+			}
+			/* store duration of txgs as its negative */
+			record.zi_duration *= -1;
+			break;
+		case 'h':
+			usage();
+			return (0);
+		case 'I':
+			/* default duration, if one hasn't yet been defined */
+			nowrites = 1;
+			if (dur_secs == 0 && dur_txg == 0)
+				record.zi_duration = 30;
+			break;
+		case 'l':
+			level = (int)strtol(optarg, &end, 10);
+			if (*end != '\0') {
+				(void) fprintf(stderr, "invalid level '%s': "
+				    "must be an integer\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 'm':
+			domount = 1;
+			break;
+		case 'p':
+			(void) strlcpy(record.zi_func, optarg,
+			    sizeof (record.zi_func));
+			break;
+		case 'q':
+			quiet = 1;
+			break;
+		case 'r':
+			range = optarg;
+			break;
+		case 's':
+			dur_secs = 1;
+			record.zi_duration = (int)strtol(optarg, &end, 10);
+			if (record.zi_duration <= 0 || *end != '\0') {
+				(void) fprintf(stderr, "invalid duration '%s': "
+				    "must be a positive integer\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 'T':
+			if (strcasecmp(optarg, "read") == 0) {
+				io_type = ZIO_TYPE_READ;
+			} else if (strcasecmp(optarg, "write") == 0) {
+				io_type = ZIO_TYPE_WRITE;
+			} else if (strcasecmp(optarg, "free") == 0) {
+				io_type = ZIO_TYPE_FREE;
+			} else if (strcasecmp(optarg, "claim") == 0) {
+				io_type = ZIO_TYPE_CLAIM;
+			} else if (strcasecmp(optarg, "all") == 0) {
+				io_type = ZIO_TYPES;
+			} else {
+				(void) fprintf(stderr, "invalid I/O type "
+				    "'%s': must be 'read', 'write', 'free', "
+				    "'claim' or 'all'\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 't':
+			if ((type = name_to_type(optarg)) == TYPE_INVAL &&
+			    !MOS_TYPE(type)) {
+				(void) fprintf(stderr, "invalid type '%s'\n",
+				    optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case 'u':
+			flags |= ZINJECT_UNLOAD_SPA;
+			break;
+		case 'L':
+			if ((label = name_to_type(optarg)) == TYPE_INVAL &&
+			    !LABEL_TYPE(type)) {
+				(void) fprintf(stderr, "invalid label type "
+				    "'%s'\n", optarg);
+				usage();
+				return (1);
+			}
+			break;
+		case ':':
+			(void) fprintf(stderr, "option -%c requires an "
+			    "operand\n", optopt);
+			usage();
+			return (1);
+		case '?':
+			(void) fprintf(stderr, "invalid option '%c'\n",
+			    optopt);
+			usage();
+			return (2);
+		}
+	}
+
+	argc -= optind;
+	argv += optind;
+
+	if (cancel != NULL) {
+		/*
+		 * '-c' is invalid with any other options.
+		 */
+		if (raw != NULL || range != NULL || type != TYPE_INVAL ||
+		    level != 0 || record.zi_func[0] != '\0' ||
+		    record.zi_duration != 0) {
+			(void) fprintf(stderr, "cancel (-c) incompatible with "
+			    "any other options\n");
+			usage();
+			return (2);
+		}
+		if (argc != 0) {
+			(void) fprintf(stderr, "extraneous argument to '-c'\n");
+			usage();
+			return (2);
+		}
+
+		if (strcmp(cancel, "all") == 0) {
+			return (cancel_all_handlers());
+		} else {
+			int id = (int)strtol(cancel, &end, 10);
+			if (*end != '\0') {
+				(void) fprintf(stderr, "invalid handle id '%s':"
+				    " must be an integer or 'all'\n", cancel);
+				usage();
+				return (1);
+			}
+			return (cancel_handler(id));
+		}
+	}
+
+	if (device != NULL) {
+		/*
+		 * Device (-d) injection uses a completely different mechanism
+		 * for doing injection, so handle it separately here.
+		 */
+		if (raw != NULL || range != NULL || type != TYPE_INVAL ||
+		    level != 0 || record.zi_func[0] != '\0' ||
+		    record.zi_duration != 0) {
+			(void) fprintf(stderr, "device (-d) incompatible with "
+			    "data error injection\n");
+			usage();
+			return (2);
+		}
+
+		if (argc != 1) {
+			(void) fprintf(stderr, "device (-d) injection requires "
+			    "a single pool name\n");
+			usage();
+			return (2);
+		}
+
+		(void) strcpy(pool, argv[0]);
+		dataset[0] = '\0';
+
+		if (error == ECKSUM) {
+			(void) fprintf(stderr, "device error type must be "
+			    "'io' or 'nxio'\n");
+			return (1);
+		}
+
+		record.zi_iotype = io_type;
+		if (translate_device(pool, device, label, &record) != 0)
+			return (1);
+		if (!error)
+			error = ENXIO;
+
+		if (action != VDEV_STATE_UNKNOWN)
+			return (perform_action(pool, &record, action));
+
+	} else if (raw != NULL) {
+		if (range != NULL || type != TYPE_INVAL || level != 0 ||
+		    record.zi_func[0] != '\0' || record.zi_duration != 0) {
+			(void) fprintf(stderr, "raw (-b) format with "
+			    "any other options\n");
+			usage();
+			return (2);
+		}
+
+		if (argc != 1) {
+			(void) fprintf(stderr, "raw (-b) format expects a "
+			    "single pool name\n");
+			usage();
+			return (2);
+		}
+
+		(void) strcpy(pool, argv[0]);
+		dataset[0] = '\0';
+
+		if (error == ENXIO) {
+			(void) fprintf(stderr, "data error type must be "
+			    "'checksum' or 'io'\n");
+			return (1);
+		}
+
+		if (translate_raw(raw, &record) != 0)
+			return (1);
+		if (!error)
+			error = EIO;
+	} else if (record.zi_func[0] != '\0') {
+		if (raw != NULL || range != NULL || type != TYPE_INVAL ||
+		    level != 0 || device != NULL || record.zi_duration != 0) {
+			(void) fprintf(stderr, "panic (-p) incompatible with "
+			    "other options\n");
+			usage();
+			return (2);
+		}
+
+		if (argc < 1 || argc > 2) {
+			(void) fprintf(stderr, "panic (-p) injection requires "
+			    "a single pool name and an optional id\n");
+			usage();
+			return (2);
+		}
+
+		(void) strcpy(pool, argv[0]);
+		if (argv[1] != NULL)
+			record.zi_type = atoi(argv[1]);
+		dataset[0] = '\0';
+	} else if (record.zi_duration != 0) {
+		if (nowrites == 0) {
+			(void) fprintf(stderr, "-s or -g meaningless "
+			    "without -I (ignore writes)\n");
+			usage();
+			return (2);
+		} else if (dur_secs && dur_txg) {
+			(void) fprintf(stderr, "choose a duration either "
+			    "in seconds (-s) or a number of txgs (-g) "
+			    "but not both\n");
+			usage();
+			return (2);
+		} else if (argc != 1) {
+			(void) fprintf(stderr, "ignore writes (-I) "
+			    "injection requires a single pool name\n");
+			usage();
+			return (2);
+		}
+
+		(void) strcpy(pool, argv[0]);
+		dataset[0] = '\0';
+	} else if (type == TYPE_INVAL) {
+		if (flags == 0) {
+			(void) fprintf(stderr, "at least one of '-b', '-d', "
+			    "'-t', '-a', '-p', '-I' or '-u' "
+			    "must be specified\n");
+			usage();
+			return (2);
+		}
+
+		if (argc == 1 && (flags & ZINJECT_UNLOAD_SPA)) {
+			(void) strcpy(pool, argv[0]);
+			dataset[0] = '\0';
+		} else if (argc != 0) {
+			(void) fprintf(stderr, "extraneous argument for "
+			    "'-f'\n");
+			usage();
+			return (2);
+		}
+
+		flags |= ZINJECT_NULL;
+	} else {
+		if (argc != 1) {
+			(void) fprintf(stderr, "missing object\n");
+			usage();
+			return (2);
+		}
+
+		if (error == ENXIO) {
+			(void) fprintf(stderr, "data error type must be "
+			    "'checksum' or 'io'\n");
+			return (1);
+		}
+
+		if (translate_record(type, argv[0], range, level, &record, pool,
+		    dataset) != 0)
+			return (1);
+		if (!error)
+			error = EIO;
+	}
+
+	/*
+	 * If this is pool-wide metadata, unmount everything.  The ioctl() will
+	 * unload the pool, so that we trigger spa-wide reopen of metadata next
+	 * time we access the pool.
+	 */
+	if (dataset[0] != '\0' && domount) {
+		if ((zhp = zfs_open(g_zfs, dataset, ZFS_TYPE_DATASET)) == NULL)
+			return (1);
+
+		if (zfs_unmount(zhp, NULL, 0) != 0)
+			return (1);
+	}
+
+	record.zi_error = error;
+
+	ret = register_handler(pool, flags, &record, quiet);
+
+	if (dataset[0] != '\0' && domount)
+		ret = (zfs_mount(zhp, NULL, 0) != 0);
+
+	libzfs_fini(g_zfs);
+
+	return (ret);
+}

cmd/zinject/zinject.h

@@ -0,0 +1,70 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef	_ZINJECT_H
+#define	_ZINJECT_H
+
+#include <sys/zfs_ioctl.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+typedef enum {
+	TYPE_DATA,		/* plain file contents		*/
+	TYPE_DNODE,		/* metadnode contents		*/
+	TYPE_MOS,		/* all MOS data			*/
+	TYPE_MOSDIR,		/* MOS object directory		*/
+	TYPE_METASLAB,		/* metaslab objects		*/
+	TYPE_CONFIG,		/* MOS config			*/
+	TYPE_BPOBJ,		/* block pointer list		*/
+	TYPE_SPACEMAP,		/* space map objects		*/
+	TYPE_ERRLOG,		/* persistent error log		*/
+	TYPE_LABEL_UBERBLOCK,	/* label specific uberblock	*/
+	TYPE_LABEL_NVLIST,	/* label specific nvlist	*/
+	TYPE_LABEL_PAD1,	/* label specific 8K pad1 area	*/
+	TYPE_LABEL_PAD2,	/* label specific 8K pad2 area	*/
+	TYPE_INVAL
+} err_type_t;
+
+#define	MOS_TYPE(t)	\
+	((t) >= TYPE_MOS && (t) < TYPE_LABEL_UBERBLOCK)
+
+#define	LABEL_TYPE(t)	\
+	((t) >= TYPE_LABEL_UBERBLOCK && (t) < TYPE_INVAL)
+
+int translate_record(err_type_t type, const char *object, const char *range,
+    int level, zinject_record_t *record, char *poolname, char *dataset);
+int translate_raw(const char *raw, zinject_record_t *record);
+int translate_device(const char *pool, const char *device,
+    err_type_t label_type, zinject_record_t *record);
+void usage(void);
+
+extern libzfs_handle_t *g_zfs;
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _ZINJECT_H */

cmd/zlook/zlook.c

@@ -0,0 +1,411 @@
+/*
+ * 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+/*
+ * This is a test program that uses ioctls to the ZFS Unit Test driver
+ * to perform readdirs or lookups using flags not normally available
+ * to user-land programs.  This allows testing of the flags'
+ * behavior outside of a complicated consumer, such as the SMB driver.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <stropts.h>
+#include <errno.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <sys/dirent.h>
+#include <sys/attr.h>
+#include <stddef.h>
+#include <fcntl.h>
+#include <string.h>
+#include <time.h>
+
+#define	_KERNEL
+
+#include <sys/fs/zut.h>
+#include <sys/extdirent.h>
+
+#undef	_KERNEL
+
+#define	MAXBUF (64 * 1024)
+#define	BIGBUF 4096
+#define	LILBUF (sizeof (dirent_t))
+
+#define	DIRENT_NAMELEN(reclen)	\
+	((reclen) - (offsetof(dirent_t, d_name[0])))
+
+static void
+usage(char *pnam)
+{
+	(void) fprintf(stderr, "Usage:\n    %s -l [-is] dir-to-look-in "
+	    "file-in-dir [xfile-on-file]\n", pnam);
+	(void) fprintf(stderr, "    %s -i [-ls] dir-to-look-in "
+	    "file-in-dir [xfile-on-file]\n", pnam);
+	(void) fprintf(stderr, "    %s -s [-il] dir-to-look-in "
+	    "file-in-dir [xfile-on-file]\n", pnam);
+	(void) fprintf(stderr, "\t    Perform a lookup\n");
+	(void) fprintf(stderr, "\t    -l == lookup\n");
+	(void) fprintf(stderr, "\t    -i == request FIGNORECASE\n");
+	(void) fprintf(stderr, "\t    -s == request stat(2) and xvattr info\n");
+	(void) fprintf(stderr, "    %s -r [-ea] [-b buffer-size-in-bytes] "
+	    "dir-to-look-in [file-in-dir]\n", pnam);
+	(void) fprintf(stderr, "    %s -e [-ra] [-b buffer-size-in-bytes] "
+	    "dir-to-look-in [file-in-dir]\n", pnam);
+	(void) fprintf(stderr, "    %s -a [-re] [-b buffer-size-in-bytes] "
+	    "dir-to-look-in [file-in-dir]\n", pnam);
+	(void) fprintf(stderr, "\t    Perform a readdir\n");
+	(void) fprintf(stderr, "\t    -r == readdir\n");
+	(void) fprintf(stderr, "\t    -e == request extended entries\n");
+	(void) fprintf(stderr, "\t    -a == request access filtering\n");
+	(void) fprintf(stderr, "\t    -b == buffer size (default 4K)\n");
+	(void) fprintf(stderr, "    %s -A path\n", pnam);
+	(void) fprintf(stderr, "\t    Look up _PC_ACCESS_FILTERING "
+	    "for path with pathconf(2)\n");
+	(void) fprintf(stderr, "    %s -E path\n", pnam);
+	(void) fprintf(stderr, "\t    Look up _PC_SATTR_EXISTS "
+	    "for path with pathconf(2)\n");
+	(void) fprintf(stderr, "    %s -S path\n", pnam);
+	(void) fprintf(stderr, "\t    Look up _PC_SATTR_EXISTS "
+	    "for path with pathconf(2)\n");
+	exit(EINVAL);
+}
+
+static void
+print_extd_entries(zut_readdir_t *r)
+{
+	struct edirent *eodp;
+	char *bufstart;
+
+	eodp = (edirent_t *)(uintptr_t)r->zr_buf;
+	bufstart = (char *)eodp;
+	while ((char *)eodp < bufstart + r->zr_bytes) {
+		char *blanks = "                ";
+		int i = 0;
+		while (i < EDIRENT_NAMELEN(eodp->ed_reclen)) {
+			if (!eodp->ed_name[i])
+				break;
+			(void) printf("%c", eodp->ed_name[i++]);
+		}
+		if (i < 16)
+			(void) printf("%.*s", 16 - i, blanks);
+		(void) printf("\t%x\n", eodp->ed_eflags);
+		eodp = (edirent_t *)((intptr_t)eodp + eodp->ed_reclen);
+	}
+}
+
+static void
+print_entries(zut_readdir_t *r)
+{
+	dirent64_t *dp;
+	char *bufstart;
+
+	dp = (dirent64_t *)(intptr_t)r->zr_buf;
+	bufstart = (char *)dp;
+	while ((char *)dp < bufstart + r->zr_bytes) {
+		int i = 0;
+		while (i < DIRENT_NAMELEN(dp->d_reclen)) {
+			if (!dp->d_name[i])
+				break;
+			(void) printf("%c", dp->d_name[i++]);
+		}
+		(void) printf("\n");
+		dp = (dirent64_t *)((intptr_t)dp + dp->d_reclen);
+	}
+}
+
+static void
+print_stats(struct stat64 *sb)
+{
+	char timebuf[512];
+
+	(void) printf("st_mode\t\t\t%04lo\n", (unsigned long)sb->st_mode);
+	(void) printf("st_ino\t\t\t%llu\n", (unsigned long long)sb->st_ino);
+	(void) printf("st_nlink\t\t%lu\n", (unsigned long)sb->st_nlink);
+	(void) printf("st_uid\t\t\t%d\n", sb->st_uid);
+	(void) printf("st_gid\t\t\t%d\n", sb->st_gid);
+	(void) printf("st_size\t\t\t%lld\n", (long long)sb->st_size);
+	(void) printf("st_blksize\t\t%ld\n", (long)sb->st_blksize);
+	(void) printf("st_blocks\t\t%lld\n", (long long)sb->st_blocks);
+
+	timebuf[0] = 0;
+	if (ctime_r(&sb->st_atime, timebuf, 512)) {
+		(void) printf("st_atime\t\t");
+		(void) printf("%s", timebuf);
+	}
+	timebuf[0] = 0;
+	if (ctime_r(&sb->st_mtime, timebuf, 512)) {
+		(void) printf("st_mtime\t\t");
+		(void) printf("%s", timebuf);
+	}
+	timebuf[0] = 0;
+	if (ctime_r(&sb->st_ctime, timebuf, 512)) {
+		(void) printf("st_ctime\t\t");
+		(void) printf("%s", timebuf);
+	}
+}
+
+static void
+print_xvs(uint64_t xvs)
+{
+	uint_t bits;
+	int idx = 0;
+
+	if (xvs == 0)
+		return;
+
+	(void) printf("-------------------\n");
+	(void) printf("Attribute bit(s) set:\n");
+	(void) printf("-------------------\n");
+
+	bits = xvs & ((1 << F_ATTR_ALL) - 1);
+	while (bits) {
+		uint_t rest = bits >> 1;
+		if (bits & 1) {
+			(void) printf("%s", attr_to_name((f_attr_t)idx));
+			if (rest)
+				(void) printf(", ");
+		}
+		idx++;
+		bits = rest;
+	}
+	(void) printf("\n");
+}
+
+int
+main(int argc, char **argv)
+{
+	zut_lookup_t lk = {0};
+	zut_readdir_t rd = {0};
+	boolean_t checking = B_FALSE;
+	boolean_t looking = B_FALSE;
+	boolean_t reading = B_FALSE;
+	boolean_t bflag = B_FALSE;
+	long rddir_bufsize = BIGBUF;
+	int error = 0;
+	int check;
+	int fd;
+	int c;
+
+	while ((c = getopt(argc, argv, "lisaerb:ASE")) != -1) {
+		switch (c) {
+		case 'l':
+			looking = B_TRUE;
+			break;
+		case 'i':
+			lk.zl_reqflags |= ZUT_IGNORECASE;
+			looking = B_TRUE;
+			break;
+		case 's':
+			lk.zl_reqflags |= ZUT_GETSTAT;
+			looking = B_TRUE;
+			break;
+		case 'a':
+			rd.zr_reqflags |= ZUT_ACCFILTER;
+			reading = B_TRUE;
+			break;
+		case 'e':
+			rd.zr_reqflags |= ZUT_EXTRDDIR;
+			reading = B_TRUE;
+			break;
+		case 'r':
+			reading = B_TRUE;
+			break;
+		case 'b':
+			reading = B_TRUE;
+			bflag = B_TRUE;
+			rddir_bufsize = strtol(optarg, NULL, 0);
+			break;
+		case 'A':
+			checking = B_TRUE;
+			check = _PC_ACCESS_FILTERING;
+			break;
+		case 'S':
+			checking = B_TRUE;
+			check = _PC_SATTR_ENABLED;
+			break;
+		case 'E':
+			checking = B_TRUE;
+			check = _PC_SATTR_EXISTS;
+			break;
+		case '?':
+		default:
+			usage(argv[0]);		/* no return */
+		}
+	}
+
+	if ((checking && looking) || (checking && reading) ||
+	    (looking && reading) || (!reading && bflag) ||
+	    (!checking && !reading && !looking))
+		usage(argv[0]);		/* no return */
+
+	if (rddir_bufsize < LILBUF || rddir_bufsize > MAXBUF) {
+		(void) fprintf(stderr, "Sorry, buffer size "
+		    "must be >= %d and less than or equal to %d bytes.\n",
+		    (int)LILBUF, MAXBUF);
+		exit(EINVAL);
+	}
+
+	if (checking) {
+		char pathbuf[MAXPATHLEN];
+		long result;
+
+		if (argc - optind < 1)
+			usage(argv[0]);		/* no return */
+		(void) strlcpy(pathbuf, argv[optind], MAXPATHLEN);
+		result = pathconf(pathbuf, check);
+		(void) printf("pathconf(2) check for %s\n", pathbuf);
+		switch (check) {
+		case _PC_SATTR_ENABLED:
+			(void) printf("System attributes ");
+			if (result != 0)
+				(void) printf("Enabled\n");
+			else
+				(void) printf("Not enabled\n");
+			break;
+		case _PC_SATTR_EXISTS:
+			(void) printf("System attributes ");
+			if (result != 0)
+				(void) printf("Exist\n");
+			else
+				(void) printf("Do not exist\n");
+			break;
+		case _PC_ACCESS_FILTERING:
+			(void) printf("Access filtering ");
+			if (result != 0)
+				(void) printf("Available\n");
+			else
+				(void) printf("Not available\n");
+			break;
+		}
+		return (result);
+	}
+
+	if ((fd = open(ZUT_DEV, O_RDONLY)) < 0) {
+		perror(ZUT_DEV);
+		return (ENXIO);
+	}
+
+	if (reading) {
+		char *buf;
+
+		if (argc - optind < 1)
+			usage(argv[0]);		/* no return */
+
+		(void) strlcpy(rd.zr_dir, argv[optind], MAXPATHLEN);
+		if (argc - optind > 1) {
+			(void) strlcpy(rd.zr_file, argv[optind + 1],
+			    MAXNAMELEN);
+			rd.zr_reqflags |= ZUT_XATTR;
+		}
+
+		if ((buf = malloc(rddir_bufsize)) == NULL) {
+			error = errno;
+			perror("malloc");
+			(void) close(fd);
+			return (error);
+		}
+
+		rd.zr_buf = (uint64_t)(uintptr_t)buf;
+		rd.zr_buflen = rddir_bufsize;
+
+		while (!rd.zr_eof) {
+			int ierr;
+
+			if ((ierr = ioctl(fd, ZUT_IOC_READDIR, &rd)) != 0) {
+				(void) fprintf(stderr,
+				    "IOCTL error: %s (%d)\n",
+				    strerror(ierr), ierr);
+				free(buf);
+				(void) close(fd);
+				return (ierr);
+			}
+			if (rd.zr_retcode) {
+				(void) fprintf(stderr,
+				    "readdir result: %s (%d)\n",
+				    strerror(rd.zr_retcode), rd.zr_retcode);
+				free(buf);
+				(void) close(fd);
+				return (rd.zr_retcode);
+			}
+			if (rd.zr_reqflags & ZUT_EXTRDDIR)
+				print_extd_entries(&rd);
+			else
+				print_entries(&rd);
+		}
+		free(buf);
+	} else {
+		int ierr;
+
+		if (argc - optind < 2)
+			usage(argv[0]);		/* no return */
+
+		(void) strlcpy(lk.zl_dir, argv[optind], MAXPATHLEN);
+		(void) strlcpy(lk.zl_file, argv[optind + 1], MAXNAMELEN);
+		if (argc - optind > 2) {
+			(void) strlcpy(lk.zl_xfile,
+			    argv[optind + 2], MAXNAMELEN);
+			lk.zl_reqflags |= ZUT_XATTR;
+		}
+
+		if ((ierr = ioctl(fd, ZUT_IOC_LOOKUP, &lk)) != 0) {
+			(void) fprintf(stderr,
+			    "IOCTL error: %s (%d)\n",
+			    strerror(ierr), ierr);
+			(void) close(fd);
+			return (ierr);
+		}
+
+		(void) printf("\nLookup of ");
+		if (lk.zl_reqflags & ZUT_XATTR) {
+			(void) printf("extended attribute \"%s\" of ",
+			    lk.zl_xfile);
+		}
+		(void) printf("file \"%s\" ", lk.zl_file);
+		(void) printf("in directory \"%s\" ", lk.zl_dir);
+		if (lk.zl_retcode) {
+			(void) printf("failed: %s (%d)\n",
+			    strerror(lk.zl_retcode), lk.zl_retcode);
+			(void) close(fd);
+			return (lk.zl_retcode);
+		}
+
+		(void) printf("succeeded.\n");
+		if (lk.zl_reqflags & ZUT_IGNORECASE) {
+			(void) printf("----------------------------\n");
+			(void) printf("dirent flags: 0x%0x\n", lk.zl_deflags);
+			(void) printf("real name: %s\n", lk.zl_real);
+		}
+		if (lk.zl_reqflags & ZUT_GETSTAT) {
+			(void) printf("----------------------------\n");
+			print_stats(&lk.zl_statbuf);
+			print_xvs(lk.zl_xvattrs);
+		}
+	}
+
+	(void) close(fd);
+	return (0);
+}

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);
+}

cmd/zpool/zpool_util.c

@@ -0,0 +1,86 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#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);
+}
+
+/*
+ * 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);
+}

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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#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);
+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 replacing, boolean_t dryrun, int argc, char **argv);
+nvlist_t *split_mirror_vdev(zpool_handle_t *zhp, char *newname,
+    nvlist_t *props, splitflags_t flags, 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 */

cmd/zstreamdump/zstreamdump.c

@@ -0,0 +1,429 @@
+/*
+ * 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 2010 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <libnvpair.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+#include <unistd.h>
+
+#include <sys/dmu.h>
+#include <sys/zfs_ioctl.h>
+#include <zfs_fletcher.h>
+
+uint64_t drr_record_count[DRR_NUMTYPES];
+uint64_t total_write_size = 0;
+uint64_t total_stream_len = 0;
+FILE *send_stream = 0;
+boolean_t do_byteswap = B_FALSE;
+boolean_t do_cksum = B_TRUE;
+#define	INITIAL_BUFLEN (1<<20)
+
+static void
+usage(void)
+{
+	(void) fprintf(stderr, "usage: zstreamdump [-v] [-C] < file\n");
+	(void) fprintf(stderr, "\t -v -- verbose\n");
+	(void) fprintf(stderr, "\t -C -- suppress checksum verification\n");
+	exit(1);
+}
+
+/*
+ * ssread - send stream read.
+ *
+ * Read while computing incremental checksum
+ */
+
+static size_t
+ssread(void *buf, size_t len, zio_cksum_t *cksum)
+{
+	size_t outlen;
+
+	if ((outlen = fread(buf, len, 1, send_stream)) == 0)
+		return (0);
+
+	if (do_cksum && cksum) {
+		if (do_byteswap)
+			fletcher_4_incremental_byteswap(buf, len, cksum);
+		else
+			fletcher_4_incremental_native(buf, len, cksum);
+	}
+	total_stream_len += len;
+	return (outlen);
+}
+
+int
+main(int argc, char *argv[])
+{
+	char *buf = malloc(INITIAL_BUFLEN);
+	dmu_replay_record_t thedrr;
+	dmu_replay_record_t *drr = &thedrr;
+	struct drr_begin *drrb = &thedrr.drr_u.drr_begin;
+	struct drr_end *drre = &thedrr.drr_u.drr_end;
+	struct drr_object *drro = &thedrr.drr_u.drr_object;
+	struct drr_freeobjects *drrfo = &thedrr.drr_u.drr_freeobjects;
+	struct drr_write *drrw = &thedrr.drr_u.drr_write;
+	struct drr_write_byref *drrwbr = &thedrr.drr_u.drr_write_byref;
+	struct drr_free *drrf = &thedrr.drr_u.drr_free;
+	struct drr_spill *drrs = &thedrr.drr_u.drr_spill;
+	char c;
+	boolean_t verbose = B_FALSE;
+	boolean_t first = B_TRUE;
+	int err;
+	zio_cksum_t zc = { 0 };
+	zio_cksum_t pcksum = { 0 };
+
+	while ((c = getopt(argc, argv, ":vC")) != -1) {
+		switch (c) {
+		case 'C':
+			do_cksum = B_FALSE;
+			break;
+		case 'v':
+			verbose = B_TRUE;
+			break;
+		case ':':
+			(void) fprintf(stderr,
+			    "missing argument for '%c' option\n", optopt);
+			usage();
+			break;
+		case '?':
+			(void) fprintf(stderr, "invalid option '%c'\n",
+			    optopt);
+			usage();
+		}
+	}
+
+	if (isatty(STDIN_FILENO)) {
+		(void) fprintf(stderr,
+		    "Error: Backup stream can not be read "
+		    "from a terminal.\n"
+		    "You must redirect standard input.\n");
+		exit(1);
+	}
+
+	send_stream = stdin;
+	pcksum = zc;
+	while (ssread(drr, sizeof (dmu_replay_record_t), &zc)) {
+
+		if (first) {
+			if (drrb->drr_magic == BSWAP_64(DMU_BACKUP_MAGIC)) {
+				do_byteswap = B_TRUE;
+				if (do_cksum) {
+					ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0);
+					/*
+					 * recalculate header checksum now
+					 * that we know it needs to be
+					 * byteswapped.
+					 */
+					fletcher_4_incremental_byteswap(drr,
+					    sizeof (dmu_replay_record_t), &zc);
+				}
+			} else if (drrb->drr_magic != DMU_BACKUP_MAGIC) {
+				(void) fprintf(stderr, "Invalid stream "
+				    "(bad magic number)\n");
+				exit(1);
+			}
+			first = B_FALSE;
+		}
+		if (do_byteswap) {
+			drr->drr_type = BSWAP_32(drr->drr_type);
+			drr->drr_payloadlen =
+			    BSWAP_32(drr->drr_payloadlen);
+		}
+
+		/*
+		 * At this point, the leading fields of the replay record
+		 * (drr_type and drr_payloadlen) have been byte-swapped if
+		 * necessary, but the rest of the data structure (the
+		 * union of type-specific structures) is still in its
+		 * original state.
+		 */
+		if (drr->drr_type >= DRR_NUMTYPES) {
+			(void) printf("INVALID record found: type 0x%x\n",
+			    drr->drr_type);
+			(void) printf("Aborting.\n");
+			exit(1);
+		}
+
+		drr_record_count[drr->drr_type]++;
+
+		switch (drr->drr_type) {
+		case DRR_BEGIN:
+			if (do_byteswap) {
+				drrb->drr_magic = BSWAP_64(drrb->drr_magic);
+				drrb->drr_versioninfo =
+				    BSWAP_64(drrb->drr_versioninfo);
+				drrb->drr_creation_time =
+				    BSWAP_64(drrb->drr_creation_time);
+				drrb->drr_type = BSWAP_32(drrb->drr_type);
+				drrb->drr_flags = BSWAP_32(drrb->drr_flags);
+				drrb->drr_toguid = BSWAP_64(drrb->drr_toguid);
+				drrb->drr_fromguid =
+				    BSWAP_64(drrb->drr_fromguid);
+			}
+
+			(void) printf("BEGIN record\n");
+			(void) printf("\thdrtype = %lld\n",
+			    DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo));
+			(void) printf("\tfeatures = %llx\n",
+			    DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo));
+			(void) printf("\tmagic = %llx\n",
+			    (u_longlong_t)drrb->drr_magic);
+			(void) printf("\tcreation_time = %llx\n",
+			    (u_longlong_t)drrb->drr_creation_time);
+			(void) printf("\ttype = %u\n", drrb->drr_type);
+			(void) printf("\tflags = 0x%x\n", drrb->drr_flags);
+			(void) printf("\ttoguid = %llx\n",
+			    (u_longlong_t)drrb->drr_toguid);
+			(void) printf("\tfromguid = %llx\n",
+			    (u_longlong_t)drrb->drr_fromguid);
+			(void) printf("\ttoname = %s\n", drrb->drr_toname);
+			if (verbose)
+				(void) printf("\n");
+
+			if ((DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo) ==
+			    DMU_COMPOUNDSTREAM) && drr->drr_payloadlen != 0) {
+				nvlist_t *nv;
+				int sz = drr->drr_payloadlen;
+
+				if (sz > 1<<20) {
+					free(buf);
+					buf = malloc(sz);
+				}
+				(void) ssread(buf, sz, &zc);
+				if (ferror(send_stream))
+					perror("fread");
+				err = nvlist_unpack(buf, sz, &nv, 0);
+				if (err)
+					perror(strerror(err));
+				nvlist_print(stdout, nv);
+				nvlist_free(nv);
+			}
+			break;
+
+		case DRR_END:
+			if (do_byteswap) {
+				drre->drr_checksum.zc_word[0] =
+				    BSWAP_64(drre->drr_checksum.zc_word[0]);
+				drre->drr_checksum.zc_word[1] =
+				    BSWAP_64(drre->drr_checksum.zc_word[1]);
+				drre->drr_checksum.zc_word[2] =
+				    BSWAP_64(drre->drr_checksum.zc_word[2]);
+				drre->drr_checksum.zc_word[3] =
+				    BSWAP_64(drre->drr_checksum.zc_word[3]);
+			}
+			/*
+			 * We compare against the *previous* checksum
+			 * value, because the stored checksum is of
+			 * everything before the DRR_END record.
+			 */
+			if (do_cksum && !ZIO_CHECKSUM_EQUAL(drre->drr_checksum,
+			    pcksum)) {
+				(void) printf("Expected checksum differs from "
+				    "checksum in stream.\n");
+				(void) printf("Expected checksum = "
+				    "%llx/%llx/%llx/%llx\n",
+				    pcksum.zc_word[0],
+				    pcksum.zc_word[1],
+				    pcksum.zc_word[2],
+				    pcksum.zc_word[3]);
+			}
+			(void) printf("END checksum = %llx/%llx/%llx/%llx\n",
+			    drre->drr_checksum.zc_word[0],
+			    drre->drr_checksum.zc_word[1],
+			    drre->drr_checksum.zc_word[2],
+			    drre->drr_checksum.zc_word[3]);
+
+			ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0);
+			break;
+
+		case DRR_OBJECT:
+			if (do_byteswap) {
+				drro->drr_object = BSWAP_64(drro->drr_object);
+				drro->drr_type = BSWAP_32(drro->drr_type);
+				drro->drr_bonustype =
+				    BSWAP_32(drro->drr_bonustype);
+				drro->drr_blksz = BSWAP_32(drro->drr_blksz);
+				drro->drr_bonuslen =
+				    BSWAP_32(drro->drr_bonuslen);
+				drro->drr_toguid = BSWAP_64(drro->drr_toguid);
+			}
+			if (verbose) {
+				(void) printf("OBJECT object = %llu type = %u "
+				    "bonustype = %u blksz = %u bonuslen = %u\n",
+				    (u_longlong_t)drro->drr_object,
+				    drro->drr_type,
+				    drro->drr_bonustype,
+				    drro->drr_blksz,
+				    drro->drr_bonuslen);
+			}
+			if (drro->drr_bonuslen > 0) {
+				(void) ssread(buf, P2ROUNDUP(drro->drr_bonuslen,
+				    8), &zc);
+			}
+			break;
+
+		case DRR_FREEOBJECTS:
+			if (do_byteswap) {
+				drrfo->drr_firstobj =
+				    BSWAP_64(drrfo->drr_firstobj);
+				drrfo->drr_numobjs =
+				    BSWAP_64(drrfo->drr_numobjs);
+				drrfo->drr_toguid = BSWAP_64(drrfo->drr_toguid);
+			}
+			if (verbose) {
+				(void) printf("FREEOBJECTS firstobj = %llu "
+				    "numobjs = %llu\n",
+				    (u_longlong_t)drrfo->drr_firstobj,
+				    (u_longlong_t)drrfo->drr_numobjs);
+			}
+			break;
+
+		case DRR_WRITE:
+			if (do_byteswap) {
+				drrw->drr_object = BSWAP_64(drrw->drr_object);
+				drrw->drr_type = BSWAP_32(drrw->drr_type);
+				drrw->drr_offset = BSWAP_64(drrw->drr_offset);
+				drrw->drr_length = BSWAP_64(drrw->drr_length);
+				drrw->drr_toguid = BSWAP_64(drrw->drr_toguid);
+				drrw->drr_key.ddk_prop =
+				    BSWAP_64(drrw->drr_key.ddk_prop);
+			}
+			if (verbose) {
+				(void) printf("WRITE object = %llu type = %u "
+				    "checksum type = %u\n"
+				    "offset = %llu length = %llu "
+				    "props = %llx\n",
+				    (u_longlong_t)drrw->drr_object,
+				    drrw->drr_type,
+				    drrw->drr_checksumtype,
+				    (u_longlong_t)drrw->drr_offset,
+				    (u_longlong_t)drrw->drr_length,
+				    (u_longlong_t)drrw->drr_key.ddk_prop);
+			}
+			(void) ssread(buf, drrw->drr_length, &zc);
+			total_write_size += drrw->drr_length;
+			break;
+
+		case DRR_WRITE_BYREF:
+			if (do_byteswap) {
+				drrwbr->drr_object =
+				    BSWAP_64(drrwbr->drr_object);
+				drrwbr->drr_offset =
+				    BSWAP_64(drrwbr->drr_offset);
+				drrwbr->drr_length =
+				    BSWAP_64(drrwbr->drr_length);
+				drrwbr->drr_toguid =
+				    BSWAP_64(drrwbr->drr_toguid);
+				drrwbr->drr_refguid =
+				    BSWAP_64(drrwbr->drr_refguid);
+				drrwbr->drr_refobject =
+				    BSWAP_64(drrwbr->drr_refobject);
+				drrwbr->drr_refoffset =
+				    BSWAP_64(drrwbr->drr_refoffset);
+				drrwbr->drr_key.ddk_prop =
+				    BSWAP_64(drrwbr->drr_key.ddk_prop);
+			}
+			if (verbose) {
+				(void) printf("WRITE_BYREF object = %llu "
+				    "checksum type = %u props = %llx\n"
+				    "offset = %llu length = %llu\n"
+				    "toguid = %llx refguid = %llx\n"
+				    "refobject = %llu refoffset = %llu\n",
+				    (u_longlong_t)drrwbr->drr_object,
+				    drrwbr->drr_checksumtype,
+				    (u_longlong_t)drrwbr->drr_key.ddk_prop,
+				    (u_longlong_t)drrwbr->drr_offset,
+				    (u_longlong_t)drrwbr->drr_length,
+				    (u_longlong_t)drrwbr->drr_toguid,
+				    (u_longlong_t)drrwbr->drr_refguid,
+				    (u_longlong_t)drrwbr->drr_refobject,
+				    (u_longlong_t)drrwbr->drr_refoffset);
+			}
+			break;
+
+		case DRR_FREE:
+			if (do_byteswap) {
+				drrf->drr_object = BSWAP_64(drrf->drr_object);
+				drrf->drr_offset = BSWAP_64(drrf->drr_offset);
+				drrf->drr_length = BSWAP_64(drrf->drr_length);
+			}
+			if (verbose) {
+				(void) printf("FREE object = %llu "
+				    "offset = %llu length = %lld\n",
+				    (u_longlong_t)drrf->drr_object,
+				    (u_longlong_t)drrf->drr_offset,
+				    (longlong_t)drrf->drr_length);
+			}
+			break;
+		case DRR_SPILL:
+			if (do_byteswap) {
+				drrs->drr_object = BSWAP_64(drrs->drr_object);
+				drrs->drr_length = BSWAP_64(drrs->drr_length);
+			}
+			if (verbose) {
+				(void) printf("SPILL block for object = %llu "
+				    "length = %llu\n", drrs->drr_object,
+				    drrs->drr_length);
+			}
+			(void) ssread(buf, drrs->drr_length, &zc);
+			break;
+		}
+		pcksum = zc;
+	}
+	free(buf);
+
+	/* Print final summary */
+
+	(void) printf("SUMMARY:\n");
+	(void) printf("\tTotal DRR_BEGIN records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_BEGIN]);
+	(void) printf("\tTotal DRR_END records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_END]);
+	(void) printf("\tTotal DRR_OBJECT records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_OBJECT]);
+	(void) printf("\tTotal DRR_FREEOBJECTS records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_FREEOBJECTS]);
+	(void) printf("\tTotal DRR_WRITE records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_WRITE]);
+	(void) printf("\tTotal DRR_FREE records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_FREE]);
+	(void) printf("\tTotal DRR_SPILL records = %lld\n",
+	    (u_longlong_t)drr_record_count[DRR_SPILL]);
+	(void) printf("\tTotal records = %lld\n",
+	    (u_longlong_t)(drr_record_count[DRR_BEGIN] +
+	    drr_record_count[DRR_OBJECT] +
+	    drr_record_count[DRR_FREEOBJECTS] +
+	    drr_record_count[DRR_WRITE] +
+	    drr_record_count[DRR_FREE] +
+	    drr_record_count[DRR_SPILL] +
+	    drr_record_count[DRR_END]));
+	(void) printf("\tTotal write size = %lld (0x%llx)\n",
+	    (u_longlong_t)total_write_size, (u_longlong_t)total_write_size);
+	(void) printf("\tTotal stream length = %lld (0x%llx)\n",
+	    (u_longlong_t)total_stream_len, (u_longlong_t)total_stream_len);
+	return (0);
+}

head/atomic.h

@@ -0,0 +1,34 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2005 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef	_ATOMIC_H
+#define	_ATOMIC_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <sys/atomic.h>
+
+#endif	/* _ATOMIC_H */

head/libintl.h

@@ -0,0 +1,125 @@
+/*
+ * 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	_LIBINTL_H
+#define	_LIBINTL_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <sys/isa_defs.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * wchar_t is a built-in type in standard C++ and as such is not
+ * defined here when using standard C++. However, the GNU compiler
+ * fixincludes utility nonetheless creates its own version of this
+ * header for use by gcc and g++. In that version it adds a redundant
+ * guard for __cplusplus. To avoid the creation of a gcc/g++ specific
+ * header we need to include the following magic comment:
+ *
+ * we must use the C++ compiler's type
+ *
+ * The above comment should not be removed or changed until GNU
+ * gcc/fixinc/inclhack.def is updated to bypass this header.
+ */
+#if !defined(__cplusplus) || (__cplusplus < 199711L && !defined(__GNUG__))
+#ifndef _WCHAR_T
+#define	_WCHAR_T
+#if defined(_LP64)
+typedef int	wchar_t;
+#else
+typedef long	wchar_t;
+#endif
+#endif	/* !_WCHAR_T */
+#endif	/* !defined(__cplusplus) ... */
+
+#define	TEXTDOMAINMAX	256
+
+#define	__GNU_GETTEXT_SUPPORTED_REVISION(m)	\
+	((((m) == 0) || ((m) == 1)) ? 1 : -1)
+
+#ifdef __STDC__
+extern char *dcgettext(const char *, const char *, const int);
+extern char *dgettext(const char *, const char *);
+extern char *gettext(const char *);
+extern char *textdomain(const char *);
+extern char *bindtextdomain(const char *, const char *);
+
+/*
+ * LI18NUX 2000 Globalization Specification Version 1.0
+ * with Amendment 2
+ */
+extern char *dcngettext(const char *, const char *,
+	const char *, unsigned long int, int);
+extern char *dngettext(const char *, const char *,
+	const char *, unsigned long int);
+extern char *ngettext(const char *, const char *, unsigned long int);
+extern char *bind_textdomain_codeset(const char *, const char *);
+
+/* Word handling functions --- requires dynamic linking */
+/* Warning: these are experimental and subject to change. */
+extern int wdinit(void);
+extern int wdchkind(wchar_t);
+extern int wdbindf(wchar_t, wchar_t, int);
+extern wchar_t *wddelim(wchar_t, wchar_t, int);
+extern wchar_t mcfiller(void);
+extern int mcwrap(void);
+
+#else
+extern char *dcgettext();
+extern char *dgettext();
+extern char *gettext();
+extern char *textdomain();
+extern char *bindtextdomain();
+
+/*
+ * LI18NUX 2000 Globalization Specification Version 1.0
+ * with Amendment 2
+ */
+extern char *dcngettext();
+extern char *dngettext();
+extern char *ngettext();
+extern char *bind_textdomain_codeset();
+
+/* Word handling functions --- requires dynamic linking */
+/* Warning: these are experimental and subject to change. */
+extern int wdinit();
+extern int wdchkind();
+extern int wdbindf();
+extern wchar_t *wddelim();
+extern wchar_t mcfiller();
+extern int mcwrap();
+
+#endif
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif /* _LIBINTL_H */

head/stdio_ext.h

@@ -0,0 +1,88 @@
+/*
+ * 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 2006 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+/*
+ * Extensions to the stdio package
+ */
+
+#ifndef _STDIO_EXT_H
+#define	_STDIO_EXT_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <stdio.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * Even though the contents of the stdio FILE structure have always been
+ * private to the stdio implementation, over the years some programs have
+ * needed to get information about a stdio stream that was not accessible
+ * through a supported interface. These programs have resorted to accessing
+ * fields of the FILE structure directly, rendering them possibly non-portable
+ * to new implementations of stdio, or more likely, preventing enhancements
+ * to stdio because those programs will break.
+ *
+ * In the 64-bit world, the FILE structure is opaque. The routines here
+ * are provided as a way to get the information that used to be retrieved
+ * directly from the FILE structure. They are based on the needs of
+ * existing programs (such as 'mh' and 'emacs'), so they may be extended
+ * as other programs are ported. Though they may still be non-portable to
+ * other operating systems, they will work from each Solaris release to
+ * the next. More portable interfaces are being developed.
+ */
+
+#define	FSETLOCKING_QUERY	0
+#define	FSETLOCKING_INTERNAL	1
+#define	FSETLOCKING_BYCALLER	2
+
+extern size_t __fbufsize(FILE *stream);
+extern int __freading(FILE *stream);
+extern int __fwriting(FILE *stream);
+extern int __freadable(FILE *stream);
+extern int __fwritable(FILE *stream);
+extern int __flbf(FILE *stream);
+extern void __fpurge(FILE *stream);
+extern size_t __fpending(FILE *stream);
+extern void _flushlbf(void);
+extern int __fsetlocking(FILE *stream, int type);
+
+/*
+ * Extended FILE enabling function.
+ */
+#if defined(_LP64) && !defined(__lint)
+#define	enable_extended_FILE_stdio(fd, act)		(0)
+#else
+extern int enable_extended_FILE_stdio(int, int);
+#endif
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _STDIO_EXT_H */

head/synch.h

@@ -0,0 +1,277 @@
+/*
+ * 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 (c) 1992, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef	_SYNCH_H
+#define	_SYNCH_H
+
+/*
+ * synch.h:
+ * definitions needed to use the thread synchronization interface
+ */
+
+#ifndef _ASM
+#include <sys/machlock.h>
+#include <sys/time_impl.h>
+#include <sys/synch.h>
+#endif /* _ASM */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef _ASM
+
+/*
+ * Semaphores
+ */
+typedef struct _sema {
+	/* this structure must be the same as sem_t in <semaphore.h> */
+	uint32_t	count;		/* semaphore count */
+	uint16_t	type;
+	uint16_t	magic;
+	upad64_t	pad1[3];	/* reserved for a mutex_t */
+	upad64_t 	pad2[2];	/* reserved for a cond_t */
+} sema_t;
+
+/*
+ * POSIX.1c Note:
+ * POSIX.1c requires that <pthread.h> define the structures pthread_mutex_t
+ * and pthread_cond_t.  These structures are identical to mutex_t (lwp_mutex_t)
+ * and cond_t (lwp_cond_t) which are defined in <synch.h>.  A nested included
+ * of <synch.h> (to allow a "#typedef mutex_t  pthread_mutex_t") would pull in
+ * non-posix symbols/constants violating the namespace restrictions.  Hence,
+ * pthread_mutex_t/pthread_cond_t have been redefined in <pthread.h> (actually
+ * in <sys/types.h>).  Any modifications done to mutex_t/lwp_mutex_t or
+ * cond_t/lwp_cond_t should also be done to pthread_mutex_t/pthread_cond_t.
+ */
+typedef lwp_mutex_t mutex_t;
+typedef lwp_cond_t cond_t;
+
+/*
+ * Readers/writer locks
+ *
+ * NOTE: The layout of this structure should be kept in sync with the layout
+ * of the correponding structure of pthread_rwlock_t in sys/types.h.
+ * Also, there is an identical structure for lwp_rwlock_t in <sys/synch.h>.
+ * Because we have to deal with C++, we cannot redefine this one as that one.
+ */
+typedef struct _rwlock {
+	int32_t		readers;	/* rwstate word */
+	uint16_t	type;
+	uint16_t	magic;
+	mutex_t		mutex;		/* used with process-shared rwlocks */
+	cond_t		readercv;	/* used only to indicate ownership */
+	cond_t		writercv;	/* used only to indicate ownership */
+} rwlock_t;
+
+#ifdef	__STDC__
+int	_lwp_mutex_lock(lwp_mutex_t *);
+int	_lwp_mutex_unlock(lwp_mutex_t *);
+int	_lwp_mutex_trylock(lwp_mutex_t *);
+int	_lwp_cond_wait(lwp_cond_t *, lwp_mutex_t *);
+int	_lwp_cond_timedwait(lwp_cond_t *, lwp_mutex_t *, timespec_t *);
+int	_lwp_cond_reltimedwait(lwp_cond_t *, lwp_mutex_t *, timespec_t *);
+int	_lwp_cond_signal(lwp_cond_t *);
+int	_lwp_cond_broadcast(lwp_cond_t *);
+int	_lwp_sema_init(lwp_sema_t *, int);
+int	_lwp_sema_wait(lwp_sema_t *);
+int	_lwp_sema_trywait(lwp_sema_t *);
+int	_lwp_sema_post(lwp_sema_t *);
+int	cond_init(cond_t *, int, void *);
+int	cond_destroy(cond_t *);
+int	cond_wait(cond_t *, mutex_t *);
+int	cond_timedwait(cond_t *, mutex_t *, const timespec_t *);
+int	cond_reltimedwait(cond_t *, mutex_t *, const timespec_t *);
+int	cond_signal(cond_t *);
+int	cond_broadcast(cond_t *);
+int	mutex_init(mutex_t *, int, void *);
+int	mutex_destroy(mutex_t *);
+int	mutex_consistent(mutex_t *);
+int	mutex_lock(mutex_t *);
+int	mutex_trylock(mutex_t *);
+int	mutex_unlock(mutex_t *);
+int	rwlock_init(rwlock_t *, int, void *);
+int	rwlock_destroy(rwlock_t *);
+int	rw_rdlock(rwlock_t *);
+int	rw_wrlock(rwlock_t *);
+int	rw_unlock(rwlock_t *);
+int	rw_tryrdlock(rwlock_t *);
+int	rw_trywrlock(rwlock_t *);
+int	sema_init(sema_t *, unsigned int, int, void *);
+int	sema_destroy(sema_t *);
+int	sema_wait(sema_t *);
+int	sema_timedwait(sema_t *, const timespec_t *);
+int	sema_reltimedwait(sema_t *, const timespec_t *);
+int	sema_post(sema_t *);
+int	sema_trywait(sema_t *);
+
+#else	/* __STDC__ */
+
+int	_lwp_mutex_lock();
+int	_lwp_mutex_unlock();
+int	_lwp_mutex_trylock();
+int	_lwp_cond_wait();
+int	_lwp_cond_timedwait();
+int	_lwp_cond_reltimedwait();
+int	_lwp_cond_signal();
+int	_lwp_cond_broadcast();
+int	_lwp_sema_init();
+int	_lwp_sema_wait();
+int	_lwp_sema_trywait();
+int	_lwp_sema_post();
+int	cond_init();
+int	cond_destroy();
+int	cond_wait();
+int	cond_timedwait();
+int	cond_reltimedwait();
+int	cond_signal();
+int	cond_broadcast();
+int	mutex_init();
+int	mutex_destroy();
+int	mutex_consistent();
+int	mutex_lock();
+int	mutex_trylock();
+int	mutex_unlock();
+int	rwlock_init();
+int	rwlock_destroy();
+int	rw_rdlock();
+int	rw_wrlock();
+int	rw_unlock();
+int	rw_tryrdlock();
+int	rw_trywrlock();
+int	sema_init();
+int	sema_destroy();
+int	sema_wait();
+int	sema_timedwait();
+int	sema_reltimedwait();
+int	sema_post();
+int	sema_trywait();
+
+#endif	/* __STDC__ */
+
+#endif /* _ASM */
+
+/* "Magic numbers" tagging synchronization object types */
+#define	MUTEX_MAGIC	_MUTEX_MAGIC
+#define	SEMA_MAGIC	_SEMA_MAGIC
+#define	COND_MAGIC	_COND_MAGIC
+#define	RWL_MAGIC	_RWL_MAGIC
+
+/*
+ * POSIX.1c Note:
+ * DEFAULTMUTEX is defined same as PTHREAD_MUTEX_INITIALIZER in <pthread.h>.
+ * DEFAULTCV is defined same as PTHREAD_COND_INITIALIZER in <pthread.h>.
+ * DEFAULTRWLOCK is defined same as PTHREAD_RWLOCK_INITIALIZER in <pthread.h>.
+ * Any changes to these macros should be reflected in <pthread.h>
+ */
+#define	DEFAULTMUTEX	\
+	{{0, 0, 0, {USYNC_THREAD}, MUTEX_MAGIC}, \
+	{{{0, 0, 0, 0, 0, 0, 0, 0}}}, 0}
+#define	SHAREDMUTEX	\
+	{{0, 0, 0, {USYNC_PROCESS}, MUTEX_MAGIC}, \
+	{{{0, 0, 0, 0, 0, 0, 0, 0}}}, 0}
+#define	RECURSIVEMUTEX	\
+	{{0, 0, 0, {USYNC_THREAD|LOCK_RECURSIVE}, MUTEX_MAGIC}, \
+	{{{0, 0, 0, 0, 0, 0, 0, 0}}}, 0}
+#define	ERRORCHECKMUTEX	\
+	{{0, 0, 0, {USYNC_THREAD|LOCK_ERRORCHECK}, MUTEX_MAGIC}, \
+	{{{0, 0, 0, 0, 0, 0, 0, 0}}}, 0}
+#define	RECURSIVE_ERRORCHECKMUTEX	\
+	{{0, 0, 0, {USYNC_THREAD|LOCK_RECURSIVE|LOCK_ERRORCHECK}, \
+	MUTEX_MAGIC}, {{{0, 0, 0, 0, 0, 0, 0, 0}}}, 0}
+#define	DEFAULTCV	\
+	{{{0, 0, 0, 0}, USYNC_THREAD, COND_MAGIC}, 0}
+#define	SHAREDCV	\
+	{{{0, 0, 0, 0}, USYNC_PROCESS, COND_MAGIC}, 0}
+#define	DEFAULTSEMA	\
+	{0, USYNC_THREAD, SEMA_MAGIC, {0, 0, 0}, {0, 0}}
+#define	SHAREDSEMA	\
+	{0, USYNC_PROCESS, SEMA_MAGIC, {0, 0, 0}, {0, 0}}
+#define	DEFAULTRWLOCK	\
+	{0, USYNC_THREAD, RWL_MAGIC, DEFAULTMUTEX, DEFAULTCV, DEFAULTCV}
+#define	SHAREDRWLOCK	\
+	{0, USYNC_PROCESS, RWL_MAGIC, SHAREDMUTEX, SHAREDCV, SHAREDCV}
+
+/*
+ * Tests on lock states.
+ */
+#define	SEMA_HELD(x)		_sema_held(x)
+#define	RW_READ_HELD(x)		_rw_read_held(x)
+#define	RW_WRITE_HELD(x)	_rw_write_held(x)
+#define	RW_LOCK_HELD(x)		(RW_READ_HELD(x) || RW_WRITE_HELD(x))
+#define	MUTEX_HELD(x)		_mutex_held(x)
+
+/*
+ * The following definitions are for assertions which can be checked
+ * statically by tools like lock_lint.  You can also define your own
+ * run-time test for each.  If you don't, we define them to 1 so that
+ * such assertions simply pass.
+ */
+#ifndef NO_LOCKS_HELD
+#define	NO_LOCKS_HELD	1
+#endif
+#ifndef NO_COMPETING_THREADS
+#define	NO_COMPETING_THREADS	1
+#endif
+
+#ifndef _ASM
+
+#ifdef	__STDC__
+
+/*
+ * The *_held() functions apply equally well to Solaris threads
+ * and to Posix threads synchronization objects, but the formal
+ * type declarations are different, so we just declare the argument
+ * to each *_held() function to be a void *, expecting that they will
+ * be called with the proper type of argument in each case.
+ */
+int _sema_held(void *);			/* sema_t or sem_t */
+int _rw_read_held(void *);		/* rwlock_t or pthread_rwlock_t */
+int _rw_write_held(void *);		/* rwlock_t or pthread_rwlock_t */
+int _mutex_held(void *);		/* mutex_t or pthread_mutex_t */
+
+#else	/* __STDC__ */
+
+int _sema_held();
+int _rw_read_held();
+int _rw_write_held();
+int _mutex_held();
+
+#endif	/* __STDC__ */
+
+/* Pause API */
+#ifdef	__STDC__
+void smt_pause(void);
+#else	/* __STDC__ */
+void smt_pause();
+#endif	/* __STDC__ */
+
+#endif /* _ASM */
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _SYNCH_H */

head/thread.h

@@ -0,0 +1,156 @@
+/*
+ * 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.
+ */
+
+#ifndef	_THREAD_H
+#define	_THREAD_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+/*
+ * thread.h:
+ * definitions needed to use the thread interface except synchronization.
+ * use <synch.h> for thread synchronization.
+ */
+
+#ifndef _ASM
+#include <sys/signal.h>
+#include <sys/time.h>
+#include <synch.h>
+#endif	/* _ASM */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef _ASM
+typedef unsigned int thread_t;
+typedef unsigned int thread_key_t;
+#endif /* _ASM */
+
+#ifndef _ASM
+#ifdef __STDC__
+
+extern int thr_create(void *, size_t, void *(*)(void *), void *, long,
+			thread_t *);
+extern int thr_join(thread_t, thread_t *, void **);
+extern int thr_setconcurrency(int);
+extern int thr_getconcurrency(void);
+extern void thr_exit(void *) __NORETURN;
+extern thread_t thr_self(void);
+
+/*
+ * the definition of thr_sigsetmask() is not strict ansi-c since sigset_t is
+ * not in the strict ansi-c name space. Hence, include the prototype for
+ * thr_sigsetmask() only if strict ansi-c conformance is not turned on.
+ */
+#if !defined(_STRICT_STDC) || defined(__EXTENSIONS__)
+extern int thr_sigsetmask(int, const sigset_t *, sigset_t *);
+#endif
+
+/*
+ * the definition of thr_stksegment() is not strict ansi-c since stack_t is
+ * not in the strict ansi-c name space. Hence, include the prototype for
+ * thr_stksegment() only if strict ansi-c conformance is not turned on.
+ */
+#if !defined(_STRICT_STDC) || defined(__EXTENSIONS__)
+extern int thr_stksegment(stack_t *);
+#endif
+
+extern int thr_main(void);
+extern int thr_kill(thread_t, int);
+extern int thr_suspend(thread_t);
+extern int thr_continue(thread_t);
+extern void thr_yield(void);
+extern int thr_setprio(thread_t, int);
+extern int thr_getprio(thread_t, int *);
+extern int thr_keycreate(thread_key_t *, void(*)(void *));
+extern int thr_keycreate_once(thread_key_t *, void(*)(void *));
+extern int thr_setspecific(thread_key_t, void *);
+extern int thr_getspecific(thread_key_t, void **);
+extern size_t thr_min_stack(void);
+
+#else /* __STDC */
+
+extern int thr_create();
+extern int thr_join();
+extern int thr_setconcurrency();
+extern int thr_getconcurrency();
+extern void thr_exit();
+extern thread_t	thr_self();
+extern int thr_sigsetmask();
+extern int thr_stksegment();
+extern int thr_main();
+extern int thr_kill();
+extern int thr_suspend();
+extern int thr_continue();
+extern void thr_yield();
+extern int thr_setprio();
+extern int thr_getprio();
+extern int thr_keycreate();
+extern int thr_keycreate_once();
+extern int thr_setspecific();
+extern int thr_getspecific();
+extern size_t thr_min_stack();
+
+#endif /* __STDC */
+#endif /* _ASM */
+
+#define	THR_MIN_STACK	thr_min_stack()
+/*
+ * thread flags (one word bit mask)
+ */
+/*
+ * POSIX.1c Note:
+ * THR_BOUND is defined same as PTHREAD_SCOPE_SYSTEM in <pthread.h>
+ * THR_DETACHED is defined same as PTHREAD_CREATE_DETACHED in <pthread.h>
+ * Any changes in these definitions should be reflected in <pthread.h>
+ */
+#define	THR_BOUND		0x00000001	/* = PTHREAD_SCOPE_SYSTEM */
+#define	THR_NEW_LWP		0x00000002
+#define	THR_DETACHED		0x00000040	/* = PTHREAD_CREATE_DETACHED */
+#define	THR_SUSPENDED		0x00000080
+#define	THR_DAEMON		0x00000100
+
+/*
+ * The key to be created by thr_keycreate_once()
+ * must be statically initialized with THR_ONCE_KEY.
+ * This must be the same as PTHREAD_ONCE_KEY_NP in <pthread.h>
+ */
+#define	THR_ONCE_KEY	(thread_key_t)(-1)
+
+/*
+ * The available register states returned by thr_getstate().
+ */
+#define	TRS_VALID	0
+#define	TRS_NONVOLATILE	1
+#define	TRS_LWPID	2
+#define	TRS_INVALID	3
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif	/* _THREAD_H */

lib/libnvpair/libnvpair.h

@@ -0,0 +1,194 @@
+/*
+ * 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 (c) 2000, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef	_LIBNVPAIR_H
+#define	_LIBNVPAIR_H
+
+#include <sys/nvpair.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <regex.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * All interfaces described in this file are private to Solaris, and
+ * are subject to change at any time and without notice.  The public
+ * nvlist/nvpair interfaces, as documented in manpage sections 3NVPAIR,
+ * are all imported from <sys/nvpair.h> included above.
+ */
+
+extern int nvpair_value_match(nvpair_t *, int, char *, char **);
+extern int nvpair_value_match_regex(nvpair_t *, int, char *, regex_t *,
+    char **);
+
+extern void nvlist_print(FILE *, nvlist_t *);
+extern void dump_nvlist(nvlist_t *, int);
+
+/*
+ * Private nvlist printing interface that allows the caller some control
+ * over output rendering (as opposed to nvlist_print and dump_nvlist).
+ *
+ * Obtain an opaque nvlist_prtctl_t cookie using nvlist_prtctl_alloc
+ * (NULL on failure);  on return the cookie is set up for default formatting
+ * and rendering.  Quote the cookie in subsequent customisation functions and
+ * then pass the cookie to nvlist_prt to render the nvlist.  Finally,
+ * use nvlist_prtctl_free to release the cookie.
+ *
+ * For all nvlist_lookup_xxx and nvlist_lookup_xxx_array functions
+ * we have a corresponding brace of functions that appoint replacement
+ * rendering functions:
+ *
+ *	extern void nvlist_prtctl_xxx(nvlist_prtctl_t,
+ *	    void (*)(nvlist_prtctl_t ctl, void *private, const char *name,
+ *	    xxxtype value))
+ *
+ *	and
+ *
+ *	extern void nvlist_prtctl_xxx_array(nvlist_prtctl_t,
+ *	    void (*)(nvlist_prtctl_t ctl, void *private, const char *name,
+ *	    xxxtype value, uint_t count))
+ *
+ * where xxxtype is the C datatype corresponding to xxx, eg int8_t for "int8"
+ * and char * for "string".  The function that is appointed to render the
+ * specified datatype receives as arguments the cookie, the nvlist
+ * member name, the value of that member (or a pointer for array function),
+ * and (for array rendering functions) a count of the number of elements.
+ */
+
+typedef struct nvlist_prtctl *nvlist_prtctl_t;	/* opaque */
+
+enum nvlist_indent_mode {
+	NVLIST_INDENT_ABS,	/* Absolute indentation */
+	NVLIST_INDENT_TABBED	/* Indent with tabstops */
+};
+
+extern nvlist_prtctl_t nvlist_prtctl_alloc(void);
+extern void nvlist_prtctl_free(nvlist_prtctl_t);
+extern void nvlist_prt(nvlist_t *, nvlist_prtctl_t);
+
+/* Output stream */
+extern void nvlist_prtctl_setdest(nvlist_prtctl_t, FILE *);
+extern FILE *nvlist_prtctl_getdest(nvlist_prtctl_t);
+
+/* Indentation mode, start indent, indent increment; default tabbed/0/1 */
+extern void nvlist_prtctl_setindent(nvlist_prtctl_t, enum nvlist_indent_mode,
+    int, int);
+extern void nvlist_prtctl_doindent(nvlist_prtctl_t, int);
+
+enum nvlist_prtctl_fmt {
+	NVLIST_FMT_MEMBER_NAME,		/* name fmt; default "%s = " */
+	NVLIST_FMT_MEMBER_POSTAMBLE,	/* after nvlist member; default "\n" */
+	NVLIST_FMT_BTWN_ARRAY		/* between array members; default " " */
+};
+
+extern void nvlist_prtctl_setfmt(nvlist_prtctl_t, enum nvlist_prtctl_fmt,
+    const char *);
+extern void nvlist_prtctl_dofmt(nvlist_prtctl_t, enum nvlist_prtctl_fmt, ...);
+
+/*
+ * Function prototypes for interfaces that appoint a new rendering function
+ * for single-valued nvlist members.
+ *
+ * A replacement function receives arguments as follows:
+ *
+ *	nvlist_prtctl_t	Print control structure; do not change preferences
+ *			for this object from a print callback function.
+ *
+ *	void *		The function-private cookie argument registered
+ *			when the replacement function was appointed.
+ *
+ *	nvlist_t *	The full nvlist that is being processed.  The
+ *			rendering function is called to render a single
+ *			member (name and value passed as below) but it may
+ *			want to reference or incorporate other aspects of
+ *			the full nvlist.
+ *
+ *	const char *	Member name to render
+ *
+ *	valtype		Value of the member to render
+ *
+ * The function must return non-zero if it has rendered output for this
+ * member, or 0 if it wants to default to standard rendering for this
+ * one member.
+ */
+
+#define	NVLIST_PRINTCTL_SVDECL(funcname, valtype) \
+    extern void funcname(nvlist_prtctl_t, \
+    int (*)(nvlist_prtctl_t, void *, nvlist_t *, const char *, valtype), \
+    void *)
+
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_boolean, int);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_boolean_value, boolean_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_byte, uchar_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_int8, int8_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_uint8, uint8_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_int16, int16_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_uint16, uint16_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_int32, int32_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_uint32, uint32_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_int64, int64_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_uint64, uint64_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_double, double);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_string, char *);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_hrtime, hrtime_t);
+NVLIST_PRINTCTL_SVDECL(nvlist_prtctlop_nvlist, nvlist_t *);
+
+#undef	NVLIST_PRINTCTL_SVDECL	/* was just for "clarity" above */
+
+/*
+ * Function prototypes for interfaces that appoint a new rendering function
+ * for array-valued nvlist members.
+ *
+ * One additional argument is taken: uint_t for the number of array elements
+ *
+ * Return values as above.
+ */
+#define	NVLIST_PRINTCTL_AVDECL(funcname, vtype) \
+    extern void funcname(nvlist_prtctl_t, \
+    int (*)(nvlist_prtctl_t, void *, nvlist_t *, const char *, vtype, uint_t), \
+    void *)
+
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_boolean_array, boolean_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_byte_array, uchar_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_int8_array, int8_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_uint8_array, uint8_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_int16_array, int16_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_uint16_array, uint16_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_int32_array, int32_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_uint32_array, uint32_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_int64_array, int64_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_uint64_array, uint64_t *);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_string_array, char **);
+NVLIST_PRINTCTL_AVDECL(nvlist_prtctlop_nvlist_array, nvlist_t **);
+
+#undef	NVLIST_PRINTCTL_AVDECL	/* was just for "clarity" above */
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _LIBNVPAIR_H */

lib/libnvpair/nvpair_alloc_system.c

@@ -0,0 +1,59 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2004 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <sys/nvpair.h>
+#include <stdlib.h>
+
+/*ARGSUSED*/
+static void *
+nv_alloc_sys(nv_alloc_t *nva, size_t size)
+{
+	return (malloc(size));
+}
+
+/*ARGSUSED*/
+static void
+nv_free_sys(nv_alloc_t *nva, void *buf, size_t size)
+{
+	free(buf);
+}
+
+const nv_alloc_ops_t system_ops_def = {
+	NULL,			/* nv_ao_init() */
+	NULL,			/* nv_ao_fini() */
+	nv_alloc_sys,		/* nv_ao_alloc() */
+	nv_free_sys,		/* nv_ao_free() */
+	NULL			/* nv_ao_reset() */
+};
+
+nv_alloc_t nv_alloc_nosleep_def = {
+	&system_ops_def,
+	NULL
+};
+
+nv_alloc_t *nv_alloc_nosleep = &nv_alloc_nosleep_def;

lib/libuutil/common/libuutil.h

@@ -0,0 +1,390 @@
+/*
+ * 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 (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#ifndef	_LIBUUTIL_H
+#define	_LIBUUTIL_H
+
+#include <sys/types.h>
+#include <stdarg.h>
+#include <stdio.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * Standard flags codes.
+ */
+#define	UU_DEFAULT		0
+
+/*
+ * Standard error codes.
+ */
+#define	UU_ERROR_NONE		0	/* no error */
+#define	UU_ERROR_INVALID_ARGUMENT 1	/* invalid argument */
+#define	UU_ERROR_UNKNOWN_FLAG	2	/* passed flag invalid */
+#define	UU_ERROR_NO_MEMORY	3	/* out of memory */
+#define	UU_ERROR_CALLBACK_FAILED 4	/* callback-initiated error */
+#define	UU_ERROR_NOT_SUPPORTED	5	/* operation not supported */
+#define	UU_ERROR_EMPTY		6	/* no value provided */
+#define	UU_ERROR_UNDERFLOW	7	/* value is too small */
+#define	UU_ERROR_OVERFLOW	8	/* value is too value */
+#define	UU_ERROR_INVALID_CHAR	9	/* value contains unexpected char */
+#define	UU_ERROR_INVALID_DIGIT	10	/* value contains digit not in base */
+
+#define	UU_ERROR_SYSTEM		99	/* underlying system error */
+#define	UU_ERROR_UNKNOWN	100	/* error status not known */
+
+/*
+ * Standard program exit codes.
+ */
+#define	UU_EXIT_OK	(*(uu_exit_ok()))
+#define	UU_EXIT_FATAL	(*(uu_exit_fatal()))
+#define	UU_EXIT_USAGE	(*(uu_exit_usage()))
+
+/*
+ * Exit status profiles.
+ */
+#define	UU_PROFILE_DEFAULT	0
+#define	UU_PROFILE_LAUNCHER	1
+
+/*
+ * Error reporting functions.
+ */
+uint32_t uu_error(void);
+const char *uu_strerror(uint32_t);
+
+/*
+ * Program notification functions.
+ */
+extern void uu_alt_exit(int);
+extern const char *uu_setpname(char *);
+extern const char *uu_getpname(void);
+/*PRINTFLIKE1*/
+extern void uu_warn(const char *, ...);
+extern void uu_vwarn(const char *, va_list);
+/*PRINTFLIKE1*/
+extern void uu_die(const char *, ...) __NORETURN;
+extern void uu_vdie(const char *, va_list) __NORETURN;
+/*PRINTFLIKE2*/
+extern void uu_xdie(int, const char *, ...) __NORETURN;
+extern void uu_vxdie(int, const char *, va_list) __NORETURN;
+
+/*
+ * Exit status functions (not to be used directly)
+ */
+extern int *uu_exit_ok(void);
+extern int *uu_exit_fatal(void);
+extern int *uu_exit_usage(void);
+
+/*
+ * string->number conversions
+ */
+extern int uu_strtoint(const char *, void *, size_t, int, int64_t, int64_t);
+extern int uu_strtouint(const char *, void *, size_t, int, uint64_t, uint64_t);
+
+/*
+ * Debug print facility functions.
+ */
+typedef struct uu_dprintf uu_dprintf_t;
+
+typedef enum {
+	UU_DPRINTF_SILENT,
+	UU_DPRINTF_FATAL,
+	UU_DPRINTF_WARNING,
+	UU_DPRINTF_NOTICE,
+	UU_DPRINTF_INFO,
+	UU_DPRINTF_DEBUG
+} uu_dprintf_severity_t;
+
+extern uu_dprintf_t *uu_dprintf_create(const char *, uu_dprintf_severity_t,
+    uint_t);
+/*PRINTFLIKE3*/
+extern void uu_dprintf(uu_dprintf_t *, uu_dprintf_severity_t,
+    const char *, ...);
+extern void uu_dprintf_destroy(uu_dprintf_t *);
+extern const char *uu_dprintf_getname(uu_dprintf_t *);
+
+/*
+ * Identifier test flags and function.
+ */
+#define	UU_NAME_DOMAIN		0x1	/* allow SUNW, or com.sun, prefix */
+#define	UU_NAME_PATH		0x2	/* allow '/'-delimited paths */
+
+int uu_check_name(const char *, uint_t);
+
+/*
+ * File creation functions.
+ */
+extern int uu_open_tmp(const char *dir, uint_t uflags);
+
+/*
+ * Convenience functions.
+ */
+#define	UU_NELEM(a)	(sizeof (a) / sizeof ((a)[0]))
+
+/*PRINTFLIKE1*/
+extern char *uu_msprintf(const char *format, ...);
+extern void *uu_zalloc(size_t);
+extern char *uu_strdup(const char *);
+extern void uu_free(void *);
+
+extern boolean_t uu_strcaseeq(const char *a, const char *b);
+extern boolean_t uu_streq(const char *a, const char *b);
+extern char *uu_strndup(const char *s, size_t n);
+extern boolean_t uu_strbw(const char *a, const char *b);
+extern void *uu_memdup(const void *buf, size_t sz);
+extern void uu_dump(FILE *out, const char *prefix, const void *buf, size_t len);
+
+/*
+ * Comparison function type definition.
+ *   Developers should be careful in their use of the _private argument. If you
+ *   break interface guarantees, you get undefined behavior.
+ */
+typedef int uu_compare_fn_t(const void *__left, const void *__right,
+    void *__private);
+
+/*
+ * Walk variant flags.
+ *   A data structure need not provide support for all variants and
+ *   combinations.  Refer to the appropriate documentation.
+ */
+#define	UU_WALK_ROBUST		0x00000001	/* walk can survive removes */
+#define	UU_WALK_REVERSE		0x00000002	/* reverse walk order */
+
+#define	UU_WALK_PREORDER	0x00000010	/* walk tree in pre-order */
+#define	UU_WALK_POSTORDER	0x00000020	/* walk tree in post-order */
+
+/*
+ * Walk callback function return codes.
+ */
+#define	UU_WALK_ERROR		-1
+#define	UU_WALK_NEXT		0
+#define	UU_WALK_DONE		1
+
+/*
+ * Walk callback function type definition.
+ */
+typedef int uu_walk_fn_t(void *_elem, void *_private);
+
+/*
+ * lists: opaque structures
+ */
+typedef struct uu_list_pool uu_list_pool_t;
+typedef struct uu_list uu_list_t;
+
+typedef struct uu_list_node {
+	uintptr_t uln_opaque[2];
+} uu_list_node_t;
+
+typedef struct uu_list_walk uu_list_walk_t;
+
+typedef uintptr_t uu_list_index_t;
+
+/*
+ * lists: interface
+ *
+ * basic usage:
+ *	typedef struct foo {
+ *		...
+ *		uu_list_node_t foo_node;
+ *		...
+ *	} foo_t;
+ *
+ *	static int
+ *	foo_compare(void *l_arg, void *r_arg, void *private)
+ *	{
+ *		foo_t *l = l_arg;
+ *		foo_t *r = r_arg;
+ *
+ *		if (... l greater than r ...)
+ *			return (1);
+ *		if (... l less than r ...)
+ *			return (-1);
+ *		return (0);
+ *	}
+ *
+ *	...
+ *		// at initialization time
+ *		foo_pool = uu_list_pool_create("foo_pool",
+ *		    sizeof (foo_t), offsetof(foo_t, foo_node), foo_compare,
+ *		    debugging? 0 : UU_AVL_POOL_DEBUG);
+ *	...
+ */
+uu_list_pool_t *uu_list_pool_create(const char *, size_t, size_t,
+    uu_compare_fn_t *, uint32_t);
+#define	UU_LIST_POOL_DEBUG	0x00000001
+
+void uu_list_pool_destroy(uu_list_pool_t *);
+
+/*
+ * usage:
+ *
+ *	foo_t *a;
+ *	a = malloc(sizeof(*a));
+ *	uu_list_node_init(a, &a->foo_list, pool);
+ *	...
+ *	uu_list_node_fini(a, &a->foo_list, pool);
+ *	free(a);
+ */
+void uu_list_node_init(void *, uu_list_node_t *, uu_list_pool_t *);
+void uu_list_node_fini(void *, uu_list_node_t *, uu_list_pool_t *);
+
+uu_list_t *uu_list_create(uu_list_pool_t *, void *_parent, uint32_t);
+#define	UU_LIST_DEBUG	0x00000001
+#define	UU_LIST_SORTED	0x00000002	/* list is sorted */
+
+void uu_list_destroy(uu_list_t *);	/* list must be empty */
+
+size_t uu_list_numnodes(uu_list_t *);
+
+void *uu_list_first(uu_list_t *);
+void *uu_list_last(uu_list_t *);
+
+void *uu_list_next(uu_list_t *, void *);
+void *uu_list_prev(uu_list_t *, void *);
+
+int uu_list_walk(uu_list_t *, uu_walk_fn_t *, void *, uint32_t);
+
+uu_list_walk_t *uu_list_walk_start(uu_list_t *, uint32_t);
+void *uu_list_walk_next(uu_list_walk_t *);
+void uu_list_walk_end(uu_list_walk_t *);
+
+void *uu_list_find(uu_list_t *, void *, void *, uu_list_index_t *);
+void uu_list_insert(uu_list_t *, void *, uu_list_index_t);
+
+void *uu_list_nearest_next(uu_list_t *, uu_list_index_t);
+void *uu_list_nearest_prev(uu_list_t *, uu_list_index_t);
+
+void *uu_list_teardown(uu_list_t *, void **);
+
+void uu_list_remove(uu_list_t *, void *);
+
+/*
+ * lists: interfaces for non-sorted lists only
+ */
+int uu_list_insert_before(uu_list_t *, void *_target, void *_elem);
+int uu_list_insert_after(uu_list_t *, void *_target, void *_elem);
+
+/*
+ * avl trees: opaque structures
+ */
+typedef struct uu_avl_pool uu_avl_pool_t;
+typedef struct uu_avl uu_avl_t;
+
+typedef struct uu_avl_node {
+#ifdef _LP64
+	uintptr_t uan_opaque[3];
+#else
+	uintptr_t uan_opaque[4];
+#endif
+} uu_avl_node_t;
+
+typedef struct uu_avl_walk uu_avl_walk_t;
+
+typedef uintptr_t uu_avl_index_t;
+
+/*
+ * avl trees: interface
+ *
+ * basic usage:
+ *	typedef struct foo {
+ *		...
+ *		uu_avl_node_t foo_node;
+ *		...
+ *	} foo_t;
+ *
+ *	static int
+ *	foo_compare(void *l_arg, void *r_arg, void *private)
+ *	{
+ *		foo_t *l = l_arg;
+ *		foo_t *r = r_arg;
+ *
+ *		if (... l greater than r ...)
+ *			return (1);
+ *		if (... l less than r ...)
+ *			return (-1);
+ *		return (0);
+ *	}
+ *
+ *	...
+ *		// at initialization time
+ *		foo_pool = uu_avl_pool_create("foo_pool",
+ *		    sizeof (foo_t), offsetof(foo_t, foo_node), foo_compare,
+ *		    debugging? 0 : UU_AVL_POOL_DEBUG);
+ *	...
+ */
+uu_avl_pool_t *uu_avl_pool_create(const char *, size_t, size_t,
+    uu_compare_fn_t *, uint32_t);
+#define	UU_AVL_POOL_DEBUG	0x00000001
+
+void uu_avl_pool_destroy(uu_avl_pool_t *);
+
+/*
+ * usage:
+ *
+ *	foo_t *a;
+ *	a = malloc(sizeof(*a));
+ *	uu_avl_node_init(a, &a->foo_avl, pool);
+ *	...
+ *	uu_avl_node_fini(a, &a->foo_avl, pool);
+ *	free(a);
+ */
+void uu_avl_node_init(void *, uu_avl_node_t *, uu_avl_pool_t *);
+void uu_avl_node_fini(void *, uu_avl_node_t *, uu_avl_pool_t *);
+
+uu_avl_t *uu_avl_create(uu_avl_pool_t *, void *_parent, uint32_t);
+#define	UU_AVL_DEBUG	0x00000001
+
+void uu_avl_destroy(uu_avl_t *);	/* list must be empty */
+
+size_t uu_avl_numnodes(uu_avl_t *);
+
+void *uu_avl_first(uu_avl_t *);
+void *uu_avl_last(uu_avl_t *);
+
+void *uu_avl_next(uu_avl_t *, void *);
+void *uu_avl_prev(uu_avl_t *, void *);
+
+int uu_avl_walk(uu_avl_t *, uu_walk_fn_t *, void *, uint32_t);
+
+uu_avl_walk_t *uu_avl_walk_start(uu_avl_t *, uint32_t);
+void *uu_avl_walk_next(uu_avl_walk_t *);
+void uu_avl_walk_end(uu_avl_walk_t *);
+
+void *uu_avl_find(uu_avl_t *, void *, void *, uu_avl_index_t *);
+void uu_avl_insert(uu_avl_t *, void *, uu_avl_index_t);
+
+void *uu_avl_nearest_next(uu_avl_t *, uu_avl_index_t);
+void *uu_avl_nearest_prev(uu_avl_t *, uu_avl_index_t);
+
+void *uu_avl_teardown(uu_avl_t *, void **);
+
+void uu_avl_remove(uu_avl_t *, void *);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _LIBUUTIL_H */

lib/libuutil/common/libuutil_common.h

@@ -0,0 +1,35 @@
+/*
+ * 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	_LIBUUTIL_COMMON_H
+#define	_LIBUUTIL_COMMON_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <libuutil.h>
+#include <libuutil_impl.h>
+
+#endif	/* _LIBUUTIL_COMMON_H */

lib/libuutil/common/libuutil_impl.h

@@ -0,0 +1,181 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2005 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef	_LIBUUTIL_IMPL_H
+#define	_LIBUUTIL_IMPL_H
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include <libuutil.h>
+#include <pthread.h>
+
+#include <sys/avl_impl.h>
+#include <sys/byteorder.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+void uu_set_error(uint_t);
+#pragma rarely_called(uu_set_error)
+
+/*PRINTFLIKE1*/
+void uu_panic(const char *format, ...);
+#pragma rarely_called(uu_panic)
+
+struct uu_dprintf {
+	char	*uud_name;
+	uu_dprintf_severity_t uud_severity;
+	uint_t	uud_flags;
+};
+
+/*
+ * For debugging purposes, libuutil keeps around linked lists of all uu_lists
+ * and uu_avls, along with pointers to their parents.  These can cause false
+ * negatives when looking for memory leaks, so we encode the pointers by
+ * storing them with swapped endianness;  this is not perfect, but it's about
+ * the best we can do without wasting a lot of space.
+ */
+#ifdef _LP64
+#define	UU_PTR_ENCODE(ptr)		BSWAP_64((uintptr_t)(void *)(ptr))
+#else
+#define	UU_PTR_ENCODE(ptr)		BSWAP_32((uintptr_t)(void *)(ptr))
+#endif
+
+#define	UU_PTR_DECODE(ptr)		((void *)UU_PTR_ENCODE(ptr))
+
+/*
+ * uu_list structures
+ */
+typedef struct uu_list_node_impl {
+	struct uu_list_node_impl *uln_next;
+	struct uu_list_node_impl *uln_prev;
+} uu_list_node_impl_t;
+
+struct uu_list_walk {
+	uu_list_walk_t	*ulw_next;
+	uu_list_walk_t	*ulw_prev;
+
+	uu_list_t	*ulw_list;
+	int8_t		ulw_dir;
+	uint8_t		ulw_robust;
+	uu_list_node_impl_t *ulw_next_result;
+};
+
+struct uu_list {
+	uintptr_t	ul_next_enc;
+	uintptr_t	ul_prev_enc;
+
+	uu_list_pool_t	*ul_pool;
+	uintptr_t	ul_parent_enc;	/* encoded parent pointer */
+	size_t		ul_offset;
+	size_t		ul_numnodes;
+	uint8_t		ul_debug;
+	uint8_t		ul_sorted;
+	uint8_t		ul_index;	/* mark for uu_list_index_ts */
+
+	uu_list_node_impl_t ul_null_node;
+	uu_list_walk_t	ul_null_walk;	/* for robust walkers */
+};
+
+#define	UU_LIST_PTR(ptr)		((uu_list_t *)UU_PTR_DECODE(ptr))
+
+#define	UU_LIST_POOL_MAXNAME	64
+
+struct uu_list_pool {
+	uu_list_pool_t	*ulp_next;
+	uu_list_pool_t	*ulp_prev;
+
+	char		ulp_name[UU_LIST_POOL_MAXNAME];
+	size_t		ulp_nodeoffset;
+	size_t		ulp_objsize;
+	uu_compare_fn_t	*ulp_cmp;
+	uint8_t		ulp_debug;
+	uint8_t		ulp_last_index;
+	pthread_mutex_t	ulp_lock;		/* protects null_list */
+	uu_list_t	ulp_null_list;
+};
+
+/*
+ * uu_avl structures
+ */
+typedef struct avl_node		uu_avl_node_impl_t;
+
+struct uu_avl_walk {
+	uu_avl_walk_t	*uaw_next;
+	uu_avl_walk_t	*uaw_prev;
+
+	uu_avl_t	*uaw_avl;
+	void		*uaw_next_result;
+	int8_t		uaw_dir;
+	uint8_t		uaw_robust;
+};
+
+struct uu_avl {
+	uintptr_t	ua_next_enc;
+	uintptr_t	ua_prev_enc;
+
+	uu_avl_pool_t	*ua_pool;
+	uintptr_t	ua_parent_enc;
+	uint8_t		ua_debug;
+	uint8_t		ua_index;	/* mark for uu_avl_index_ts */
+
+	struct avl_tree	ua_tree;
+	uu_avl_walk_t	ua_null_walk;
+};
+
+#define	UU_AVL_PTR(x)		((uu_avl_t *)UU_PTR_DECODE(x))
+
+#define	UU_AVL_POOL_MAXNAME	64
+
+struct uu_avl_pool {
+	uu_avl_pool_t	*uap_next;
+	uu_avl_pool_t	*uap_prev;
+
+	char		uap_name[UU_AVL_POOL_MAXNAME];
+	size_t		uap_nodeoffset;
+	size_t		uap_objsize;
+	uu_compare_fn_t	*uap_cmp;
+	uint8_t		uap_debug;
+	uint8_t		uap_last_index;
+	pthread_mutex_t	uap_lock;		/* protects null_avl */
+	uu_avl_t	uap_null_avl;
+};
+
+/*
+ * atfork() handlers
+ */
+void uu_avl_lockup(void);
+void uu_avl_release(void);
+
+void uu_list_lockup(void);
+void uu_list_release(void);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _LIBUUTIL_IMPL_H */

lib/libuutil/common/uu_alloc.c

@@ -0,0 +1,135 @@
+/*
+ * 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 (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#include "libuutil_common.h"
+
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+void *
+uu_zalloc(size_t n)
+{
+	void *p = malloc(n);
+
+	if (p == NULL) {
+		uu_set_error(UU_ERROR_SYSTEM);
+		return (NULL);
+	}
+
+	(void) memset(p, 0, n);
+
+	return (p);
+}
+
+void
+uu_free(void *p)
+{
+	free(p);
+}
+
+char *
+uu_strdup(const char *str)
+{
+	char *buf = NULL;
+
+	if (str != NULL) {
+		size_t sz;
+
+		sz = strlen(str) + 1;
+		buf = uu_zalloc(sz);
+		if (buf != NULL)
+			(void) memcpy(buf, str, sz);
+	}
+	return (buf);
+}
+
+/*
+ * Duplicate up to n bytes of a string.  Kind of sort of like
+ * strdup(strlcpy(s, n)).
+ */
+char *
+uu_strndup(const char *s, size_t n)
+{
+	size_t len;
+	char *p;
+
+	len = strnlen(s, n);
+	p = uu_zalloc(len + 1);
+	if (p == NULL)
+		return (NULL);
+
+	if (len > 0)
+		(void) memcpy(p, s, len);
+	p[len] = '\0';
+
+	return (p);
+}
+
+/*
+ * Duplicate a block of memory.  Combines malloc with memcpy, much as
+ * strdup combines malloc, strlen, and strcpy.
+ */
+void *
+uu_memdup(const void *buf, size_t sz)
+{
+	void *p;
+
+	p = uu_zalloc(sz);
+	if (p == NULL)
+		return (NULL);
+	(void) memcpy(p, buf, sz);
+	return (p);
+}
+
+char *
+uu_msprintf(const char *format, ...)
+{
+	va_list args;
+	char attic[1];
+	uint_t M, m;
+	char *b;
+
+	va_start(args, format);
+	M = vsnprintf(attic, 1, format, args);
+	va_end(args);
+
+	for (;;) {
+		m = M;
+		if ((b = uu_zalloc(m + 1)) == NULL)
+			return (NULL);
+
+		va_start(args, format);
+		M = vsnprintf(b, m + 1, format, args);
+		va_end(args);
+
+		if (M == m)
+			break;		/* sizes match */
+
+		uu_free(b);
+	}
+
+	return (b);
+}

lib/libuutil/common/uu_avl.c

@@ -0,0 +1,569 @@
+/*
+ * 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.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/avl.h>
+
+static uu_avl_pool_t	uu_null_apool = { &uu_null_apool, &uu_null_apool };
+static pthread_mutex_t	uu_apool_list_lock = PTHREAD_MUTEX_INITIALIZER;
+
+/*
+ * The index mark change on every insert and delete, to catch stale
+ * references.
+ *
+ * We leave the low bit alone, since the avl code uses it.
+ */
+#define	INDEX_MAX		(sizeof (uintptr_t) - 2)
+#define	INDEX_NEXT(m)		(((m) == INDEX_MAX)? 2 : ((m) + 2) & INDEX_MAX)
+
+#define	INDEX_DECODE(i)		((i) & ~INDEX_MAX)
+#define	INDEX_ENCODE(p, n)	(((n) & ~INDEX_MAX) | (p)->ua_index)
+#define	INDEX_VALID(p, i)	(((i) & INDEX_MAX) == (p)->ua_index)
+#define	INDEX_CHECK(i)		(((i) & INDEX_MAX) != 0)
+
+/*
+ * When an element is inactive (not in a tree), we keep a marked pointer to
+ * its containing pool in its first word, and a NULL pointer in its second.
+ *
+ * On insert, we use these to verify that it comes from the correct pool.
+ */
+#define	NODE_ARRAY(p, n)	((uintptr_t *)((uintptr_t)(n) + \
+				    (pp)->uap_nodeoffset))
+
+#define	POOL_TO_MARKER(pp) (((uintptr_t)(pp) | 1))
+
+#define	DEAD_MARKER		0xc4
+
+uu_avl_pool_t *
+uu_avl_pool_create(const char *name, size_t objsize, size_t nodeoffset,
+    uu_compare_fn_t *compare_func, uint32_t flags)
+{
+	uu_avl_pool_t *pp, *next, *prev;
+
+	if (name == NULL ||
+	    uu_check_name(name, UU_NAME_DOMAIN) == -1 ||
+	    nodeoffset + sizeof (uu_avl_node_t) > objsize ||
+	    compare_func == NULL) {
+		uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+		return (NULL);
+	}
+
+	if (flags & ~UU_AVL_POOL_DEBUG) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	pp = uu_zalloc(sizeof (uu_avl_pool_t));
+	if (pp == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	(void) strlcpy(pp->uap_name, name, sizeof (pp->uap_name));
+	pp->uap_nodeoffset = nodeoffset;
+	pp->uap_objsize = objsize;
+	pp->uap_cmp = compare_func;
+	if (flags & UU_AVL_POOL_DEBUG)
+		pp->uap_debug = 1;
+	pp->uap_last_index = 0;
+
+	(void) pthread_mutex_init(&pp->uap_lock, NULL);
+
+	pp->uap_null_avl.ua_next_enc = UU_PTR_ENCODE(&pp->uap_null_avl);
+	pp->uap_null_avl.ua_prev_enc = UU_PTR_ENCODE(&pp->uap_null_avl);
+
+	(void) pthread_mutex_lock(&uu_apool_list_lock);
+	pp->uap_next = next = &uu_null_apool;
+	pp->uap_prev = prev = next->uap_prev;
+	next->uap_prev = pp;
+	prev->uap_next = pp;
+	(void) pthread_mutex_unlock(&uu_apool_list_lock);
+
+	return (pp);
+}
+
+void
+uu_avl_pool_destroy(uu_avl_pool_t *pp)
+{
+	if (pp->uap_debug) {
+		if (pp->uap_null_avl.ua_next_enc !=
+		    UU_PTR_ENCODE(&pp->uap_null_avl) ||
+		    pp->uap_null_avl.ua_prev_enc !=
+		    UU_PTR_ENCODE(&pp->uap_null_avl)) {
+			uu_panic("uu_avl_pool_destroy: Pool \"%.*s\" (%p) has "
+			    "outstanding avls, or is corrupt.\n",
+			    (int)sizeof (pp->uap_name), pp->uap_name,
+			    (void *)pp);
+		}
+	}
+	(void) pthread_mutex_lock(&uu_apool_list_lock);
+	pp->uap_next->uap_prev = pp->uap_prev;
+	pp->uap_prev->uap_next = pp->uap_next;
+	(void) pthread_mutex_unlock(&uu_apool_list_lock);
+	pp->uap_prev = NULL;
+	pp->uap_next = NULL;
+	uu_free(pp);
+}
+
+void
+uu_avl_node_init(void *base, uu_avl_node_t *np, uu_avl_pool_t *pp)
+{
+	uintptr_t *na = (uintptr_t *)np;
+
+	if (pp->uap_debug) {
+		uintptr_t offset = (uintptr_t)np - (uintptr_t)base;
+		if (offset + sizeof (*np) > pp->uap_objsize) {
+			uu_panic("uu_avl_node_init(%p, %p, %p (\"%s\")): "
+			    "offset %ld doesn't fit in object (size %ld)\n",
+			    base, (void *)np, (void *)pp, pp->uap_name,
+			    (long)offset, (long)pp->uap_objsize);
+		}
+		if (offset != pp->uap_nodeoffset) {
+			uu_panic("uu_avl_node_init(%p, %p, %p (\"%s\")): "
+			    "offset %ld doesn't match pool's offset (%ld)\n",
+			    base, (void *)np, (void *)pp, pp->uap_name,
+			    (long)offset, (long)pp->uap_objsize);
+		}
+	}
+
+	na[0] = POOL_TO_MARKER(pp);
+	na[1] = 0;
+}
+
+void
+uu_avl_node_fini(void *base, uu_avl_node_t *np, uu_avl_pool_t *pp)
+{
+	uintptr_t *na = (uintptr_t *)np;
+
+	if (pp->uap_debug) {
+		if (na[0] == DEAD_MARKER && na[1] == DEAD_MARKER) {
+			uu_panic("uu_avl_node_fini(%p, %p, %p (\"%s\")): "
+			    "node already finied\n",
+			    base, (void *)np, (void *)pp, pp->uap_name);
+		}
+		if (na[0] != POOL_TO_MARKER(pp) || na[1] != 0) {
+			uu_panic("uu_avl_node_fini(%p, %p, %p (\"%s\")): "
+			    "node corrupt, in tree, or in different pool\n",
+			    base, (void *)np, (void *)pp, pp->uap_name);
+		}
+	}
+
+	na[0] = DEAD_MARKER;
+	na[1] = DEAD_MARKER;
+	na[2] = DEAD_MARKER;
+}
+
+struct uu_avl_node_compare_info {
+	uu_compare_fn_t	*ac_compare;
+	void		*ac_private;
+	void		*ac_right;
+	void		*ac_found;
+};
+
+static int
+uu_avl_node_compare(const void *l, const void *r)
+{
+	struct uu_avl_node_compare_info *info =
+	    (struct uu_avl_node_compare_info *)l;
+
+	int res = info->ac_compare(r, info->ac_right, info->ac_private);
+
+	if (res == 0) {
+		if (info->ac_found == NULL)
+			info->ac_found = (void *)r;
+		return (-1);
+	}
+	if (res < 0)
+		return (1);
+	return (-1);
+}
+
+uu_avl_t *
+uu_avl_create(uu_avl_pool_t *pp, void *parent, uint32_t flags)
+{
+	uu_avl_t *ap, *next, *prev;
+
+	if (flags & ~UU_AVL_DEBUG) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	ap = uu_zalloc(sizeof (*ap));
+	if (ap == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	ap->ua_pool = pp;
+	ap->ua_parent_enc = UU_PTR_ENCODE(parent);
+	ap->ua_debug = pp->uap_debug || (flags & UU_AVL_DEBUG);
+	ap->ua_index = (pp->uap_last_index = INDEX_NEXT(pp->uap_last_index));
+
+	avl_create(&ap->ua_tree, &uu_avl_node_compare, pp->uap_objsize,
+	    pp->uap_nodeoffset);
+
+	ap->ua_null_walk.uaw_next = &ap->ua_null_walk;
+	ap->ua_null_walk.uaw_prev = &ap->ua_null_walk;
+
+	(void) pthread_mutex_lock(&pp->uap_lock);
+	next = &pp->uap_null_avl;
+	prev = UU_PTR_DECODE(next->ua_prev_enc);
+	ap->ua_next_enc = UU_PTR_ENCODE(next);
+	ap->ua_prev_enc = UU_PTR_ENCODE(prev);
+	next->ua_prev_enc = UU_PTR_ENCODE(ap);
+	prev->ua_next_enc = UU_PTR_ENCODE(ap);
+	(void) pthread_mutex_unlock(&pp->uap_lock);
+
+	return (ap);
+}
+
+void
+uu_avl_destroy(uu_avl_t *ap)
+{
+	uu_avl_pool_t *pp = ap->ua_pool;
+
+	if (ap->ua_debug) {
+		if (avl_numnodes(&ap->ua_tree) != 0) {
+			uu_panic("uu_avl_destroy(%p): tree not empty\n",
+			    (void *)ap);
+		}
+		if (ap->ua_null_walk.uaw_next != &ap->ua_null_walk ||
+		    ap->ua_null_walk.uaw_prev != &ap->ua_null_walk) {
+			uu_panic("uu_avl_destroy(%p):  outstanding walkers\n",
+			    (void *)ap);
+		}
+	}
+	(void) pthread_mutex_lock(&pp->uap_lock);
+	UU_AVL_PTR(ap->ua_next_enc)->ua_prev_enc = ap->ua_prev_enc;
+	UU_AVL_PTR(ap->ua_prev_enc)->ua_next_enc = ap->ua_next_enc;
+	(void) pthread_mutex_unlock(&pp->uap_lock);
+	ap->ua_prev_enc = UU_PTR_ENCODE(NULL);
+	ap->ua_next_enc = UU_PTR_ENCODE(NULL);
+
+	ap->ua_pool = NULL;
+	avl_destroy(&ap->ua_tree);
+
+	uu_free(ap);
+}
+
+size_t
+uu_avl_numnodes(uu_avl_t *ap)
+{
+	return (avl_numnodes(&ap->ua_tree));
+}
+
+void *
+uu_avl_first(uu_avl_t *ap)
+{
+	return (avl_first(&ap->ua_tree));
+}
+
+void *
+uu_avl_last(uu_avl_t *ap)
+{
+	return (avl_last(&ap->ua_tree));
+}
+
+void *
+uu_avl_next(uu_avl_t *ap, void *node)
+{
+	return (AVL_NEXT(&ap->ua_tree, node));
+}
+
+void *
+uu_avl_prev(uu_avl_t *ap, void *node)
+{
+	return (AVL_PREV(&ap->ua_tree, node));
+}
+
+static void
+_avl_walk_init(uu_avl_walk_t *wp, uu_avl_t *ap, uint32_t flags)
+{
+	uu_avl_walk_t *next, *prev;
+
+	int robust = (flags & UU_WALK_ROBUST);
+	int direction = (flags & UU_WALK_REVERSE)? -1 : 1;
+
+	(void) memset(wp, 0, sizeof (*wp));
+	wp->uaw_avl = ap;
+	wp->uaw_robust = robust;
+	wp->uaw_dir = direction;
+
+	if (direction > 0)
+		wp->uaw_next_result = avl_first(&ap->ua_tree);
+	else
+		wp->uaw_next_result = avl_last(&ap->ua_tree);
+
+	if (ap->ua_debug || robust) {
+		wp->uaw_next = next = &ap->ua_null_walk;
+		wp->uaw_prev = prev = next->uaw_prev;
+		next->uaw_prev = wp;
+		prev->uaw_next = wp;
+	}
+}
+
+static void *
+_avl_walk_advance(uu_avl_walk_t *wp, uu_avl_t *ap)
+{
+	void *np = wp->uaw_next_result;
+
+	avl_tree_t *t = &ap->ua_tree;
+
+	if (np == NULL)
+		return (NULL);
+
+	wp->uaw_next_result = (wp->uaw_dir > 0)? AVL_NEXT(t, np) :
+	    AVL_PREV(t, np);
+
+	return (np);
+}
+
+static void
+_avl_walk_fini(uu_avl_walk_t *wp)
+{
+	if (wp->uaw_next != NULL) {
+		wp->uaw_next->uaw_prev = wp->uaw_prev;
+		wp->uaw_prev->uaw_next = wp->uaw_next;
+		wp->uaw_next = NULL;
+		wp->uaw_prev = NULL;
+	}
+	wp->uaw_avl = NULL;
+	wp->uaw_next_result = NULL;
+}
+
+uu_avl_walk_t *
+uu_avl_walk_start(uu_avl_t *ap, uint32_t flags)
+{
+	uu_avl_walk_t *wp;
+
+	if (flags & ~(UU_WALK_ROBUST | UU_WALK_REVERSE)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	wp = uu_zalloc(sizeof (*wp));
+	if (wp == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	_avl_walk_init(wp, ap, flags);
+	return (wp);
+}
+
+void *
+uu_avl_walk_next(uu_avl_walk_t *wp)
+{
+	return (_avl_walk_advance(wp, wp->uaw_avl));
+}
+
+void
+uu_avl_walk_end(uu_avl_walk_t *wp)
+{
+	_avl_walk_fini(wp);
+	uu_free(wp);
+}
+
+int
+uu_avl_walk(uu_avl_t *ap, uu_walk_fn_t *func, void *private, uint32_t flags)
+{
+	void *e;
+	uu_avl_walk_t my_walk;
+
+	int status = UU_WALK_NEXT;
+
+	if (flags & ~(UU_WALK_ROBUST | UU_WALK_REVERSE)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (-1);
+	}
+
+	_avl_walk_init(&my_walk, ap, flags);
+	while (status == UU_WALK_NEXT &&
+	    (e = _avl_walk_advance(&my_walk, ap)) != NULL)
+		status = (*func)(e, private);
+	_avl_walk_fini(&my_walk);
+
+	if (status >= 0)
+		return (0);
+	uu_set_error(UU_ERROR_CALLBACK_FAILED);
+	return (-1);
+}
+
+void
+uu_avl_remove(uu_avl_t *ap, void *elem)
+{
+	uu_avl_walk_t *wp;
+	uu_avl_pool_t *pp = ap->ua_pool;
+	uintptr_t *na = NODE_ARRAY(pp, elem);
+
+	if (ap->ua_debug) {
+		/*
+		 * invalidate outstanding uu_avl_index_ts.
+		 */
+		ap->ua_index = INDEX_NEXT(ap->ua_index);
+	}
+
+	/*
+	 * Robust walkers most be advanced, if we are removing the node
+	 * they are currently using.  In debug mode, non-robust walkers
+	 * are also on the walker list.
+	 */
+	for (wp = ap->ua_null_walk.uaw_next; wp != &ap->ua_null_walk;
+	    wp = wp->uaw_next) {
+		if (wp->uaw_robust) {
+			if (elem == wp->uaw_next_result)
+				(void) _avl_walk_advance(wp, ap);
+		} else if (wp->uaw_next_result != NULL) {
+			uu_panic("uu_avl_remove(%p, %p): active non-robust "
+			    "walker\n", (void *)ap, elem);
+		}
+	}
+
+	avl_remove(&ap->ua_tree, elem);
+
+	na[0] = POOL_TO_MARKER(pp);
+	na[1] = 0;
+}
+
+void *
+uu_avl_teardown(uu_avl_t *ap, void **cookie)
+{
+	void *elem = avl_destroy_nodes(&ap->ua_tree, cookie);
+
+	if (elem != NULL) {
+		uu_avl_pool_t *pp = ap->ua_pool;
+		uintptr_t *na = NODE_ARRAY(pp, elem);
+
+		na[0] = POOL_TO_MARKER(pp);
+		na[1] = 0;
+	}
+	return (elem);
+}
+
+void *
+uu_avl_find(uu_avl_t *ap, void *elem, void *private, uu_avl_index_t *out)
+{
+	struct uu_avl_node_compare_info info;
+	void *result;
+
+	info.ac_compare = ap->ua_pool->uap_cmp;
+	info.ac_private = private;
+	info.ac_right = elem;
+	info.ac_found = NULL;
+
+	result = avl_find(&ap->ua_tree, &info, out);
+	if (out != NULL)
+		*out = INDEX_ENCODE(ap, *out);
+
+	if (ap->ua_debug && result != NULL)
+		uu_panic("uu_avl_find: internal error: avl_find succeeded\n");
+
+	return (info.ac_found);
+}
+
+void
+uu_avl_insert(uu_avl_t *ap, void *elem, uu_avl_index_t idx)
+{
+	if (ap->ua_debug) {
+		uu_avl_pool_t *pp = ap->ua_pool;
+		uintptr_t *na = NODE_ARRAY(pp, elem);
+
+		if (na[1] != 0)
+			uu_panic("uu_avl_insert(%p, %p, %p): node already "
+			    "in tree, or corrupt\n",
+			    (void *)ap, elem, (void *)idx);
+		if (na[0] == 0)
+			uu_panic("uu_avl_insert(%p, %p, %p): node not "
+			    "initialized\n",
+			    (void *)ap, elem, (void *)idx);
+		if (na[0] != POOL_TO_MARKER(pp))
+			uu_panic("uu_avl_insert(%p, %p, %p): node from "
+			    "other pool, or corrupt\n",
+			    (void *)ap, elem, (void *)idx);
+
+		if (!INDEX_VALID(ap, idx))
+			uu_panic("uu_avl_insert(%p, %p, %p): %s\n",
+			    (void *)ap, elem, (void *)idx,
+			    INDEX_CHECK(idx)? "outdated index" :
+			    "invalid index");
+
+		/*
+		 * invalidate outstanding uu_avl_index_ts.
+		 */
+		ap->ua_index = INDEX_NEXT(ap->ua_index);
+	}
+	avl_insert(&ap->ua_tree, elem, INDEX_DECODE(idx));
+}
+
+void *
+uu_avl_nearest_next(uu_avl_t *ap, uu_avl_index_t idx)
+{
+	if (ap->ua_debug && !INDEX_VALID(ap, idx))
+		uu_panic("uu_avl_nearest_next(%p, %p): %s\n",
+		    (void *)ap, (void *)idx, INDEX_CHECK(idx)?
+		    "outdated index" : "invalid index");
+	return (avl_nearest(&ap->ua_tree, INDEX_DECODE(idx), AVL_AFTER));
+}
+
+void *
+uu_avl_nearest_prev(uu_avl_t *ap, uu_avl_index_t idx)
+{
+	if (ap->ua_debug && !INDEX_VALID(ap, idx))
+		uu_panic("uu_avl_nearest_prev(%p, %p): %s\n",
+		    (void *)ap, (void *)idx, INDEX_CHECK(idx)?
+		    "outdated index" : "invalid index");
+	return (avl_nearest(&ap->ua_tree, INDEX_DECODE(idx), AVL_BEFORE));
+}
+
+/*
+ * called from uu_lockup() and uu_release(), as part of our fork1()-safety.
+ */
+void
+uu_avl_lockup(void)
+{
+	uu_avl_pool_t *pp;
+
+	(void) pthread_mutex_lock(&uu_apool_list_lock);
+	for (pp = uu_null_apool.uap_next; pp != &uu_null_apool;
+	    pp = pp->uap_next)
+		(void) pthread_mutex_lock(&pp->uap_lock);
+}
+
+void
+uu_avl_release(void)
+{
+	uu_avl_pool_t *pp;
+
+	for (pp = uu_null_apool.uap_next; pp != &uu_null_apool;
+	    pp = pp->uap_next)
+		(void) pthread_mutex_unlock(&pp->uap_lock);
+	(void) pthread_mutex_unlock(&uu_apool_list_lock);
+}

lib/libuutil/common/uu_dprintf.c

@@ -0,0 +1,128 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2004 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <errno.h>
+#include <libintl.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+
+#define	FACILITY_FMT	"%s (%s): "
+
+#if !defined(TEXT_DOMAIN)
+#define	TEXT_DOMAIN "SYS_TEST"
+#endif
+
+static const char *
+strseverity(uu_dprintf_severity_t severity)
+{
+	switch (severity) {
+	case UU_DPRINTF_SILENT:
+		return (dgettext(TEXT_DOMAIN, "silent"));
+	case UU_DPRINTF_FATAL:
+		return (dgettext(TEXT_DOMAIN, "FATAL"));
+	case UU_DPRINTF_WARNING:
+		return (dgettext(TEXT_DOMAIN, "WARNING"));
+	case UU_DPRINTF_NOTICE:
+		return (dgettext(TEXT_DOMAIN, "note"));
+	case UU_DPRINTF_INFO:
+		return (dgettext(TEXT_DOMAIN, "info"));
+	case UU_DPRINTF_DEBUG:
+		return (dgettext(TEXT_DOMAIN, "debug"));
+	default:
+		return (dgettext(TEXT_DOMAIN, "unspecified"));
+	}
+}
+
+uu_dprintf_t *
+uu_dprintf_create(const char *name, uu_dprintf_severity_t severity,
+    uint_t flags)
+{
+	uu_dprintf_t *D;
+
+	if (uu_check_name(name, UU_NAME_DOMAIN) == -1) {
+		uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+		return (NULL);
+	}
+
+	if ((D = uu_zalloc(sizeof (uu_dprintf_t))) == NULL)
+		return (NULL);
+
+	if (name != NULL) {
+		D->uud_name = strdup(name);
+		if (D->uud_name == NULL) {
+			uu_free(D);
+			return (NULL);
+		}
+	} else {
+		D->uud_name = NULL;
+	}
+
+	D->uud_severity = severity;
+	D->uud_flags = flags;
+
+	return (D);
+}
+
+/*PRINTFLIKE3*/
+void
+uu_dprintf(uu_dprintf_t *D, uu_dprintf_severity_t severity,
+    const char *format, ...)
+{
+	va_list alist;
+
+	/* XXX Assert that severity is not UU_DPRINTF_SILENT. */
+
+	if (severity > D->uud_severity)
+		return;
+
+	(void) fprintf(stderr, FACILITY_FMT, D->uud_name,
+	    strseverity(severity));
+
+	va_start(alist, format);
+	(void) vfprintf(stderr, format, alist);
+	va_end(alist);
+}
+
+void
+uu_dprintf_destroy(uu_dprintf_t *D)
+{
+	if (D->uud_name)
+		free(D->uud_name);
+
+	uu_free(D);
+}
+
+const char *
+uu_dprintf_getname(uu_dprintf_t *D)
+{
+	return (D->uud_name);
+}

lib/libuutil/common/uu_ident.c

@@ -0,0 +1,122 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2004 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <string.h>
+
+/*
+ * We require names of the form:
+ *	[provider,]identifier[/[provider,]identifier]...
+ *
+ * Where provider is either a stock symbol (SUNW) or a java-style reversed
+ * domain name (com.sun).
+ *
+ * Both providers and identifiers must start with a letter, and may
+ * only contain alphanumerics, dashes, and underlines.  Providers
+ * may also contain periods.
+ *
+ * Note that we do _not_ use the macros in <ctype.h>, since they are affected
+ * by the current locale settings.
+ */
+
+#define	IS_ALPHA(c) \
+	(((c) >= 'a' && (c) <= 'z') || ((c) >= 'A' && (c) <= 'Z'))
+
+#define	IS_DIGIT(c) \
+	((c) >= '0' && (c) <= '9')
+
+static int
+is_valid_ident(const char *s, const char *e, int allowdot)
+{
+	char c;
+
+	if (s >= e)
+		return (0);		/* name is empty */
+
+	c = *s++;
+	if (!IS_ALPHA(c))
+		return (0);		/* does not start with letter */
+
+	while (s < e && (c = *s++) != 0) {
+		if (IS_ALPHA(c) || IS_DIGIT(c) || c == '-' || c == '_' ||
+		    (allowdot && c == '.'))
+			continue;
+		return (0);		/* invalid character */
+	}
+	return (1);
+}
+
+static int
+is_valid_component(const char *b, const char *e, uint_t flags)
+{
+	char *sp;
+
+	if (flags & UU_NAME_DOMAIN) {
+		sp = strchr(b, ',');
+		if (sp != NULL && sp < e) {
+			if (!is_valid_ident(b, sp, 1))
+				return (0);
+			b = sp + 1;
+		}
+	}
+
+	return (is_valid_ident(b, e, 0));
+}
+
+int
+uu_check_name(const char *name, uint_t flags)
+{
+	const char *end = name + strlen(name);
+	const char *p;
+
+	if (flags & ~(UU_NAME_DOMAIN | UU_NAME_PATH)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (-1);
+	}
+
+	if (!(flags & UU_NAME_PATH)) {
+		if (!is_valid_component(name, end, flags))
+			goto bad;
+		return (0);
+	}
+
+	while ((p = strchr(name, '/')) != NULL) {
+		if (!is_valid_component(name, p - 1, flags))
+			goto bad;
+		name = p + 1;
+	}
+	if (!is_valid_component(name, end, flags))
+		goto bad;
+
+	return (0);
+
+bad:
+	uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+	return (-1);
+}

lib/libuutil/common/uu_list.c

@@ -0,0 +1,721 @@
+/*
+ * 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.
+ *
+ * Copyright 2011 Jason King.  All rights reserved.
+ */
+
+#include "libuutil_common.h"
+
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/time.h>
+
+#define	ELEM_TO_NODE(lp, e) \
+	((uu_list_node_impl_t *)((uintptr_t)(e) + (lp)->ul_offset))
+
+#define	NODE_TO_ELEM(lp, n) \
+	((void *)((uintptr_t)(n) - (lp)->ul_offset))
+
+/*
+ * uu_list_index_ts define a location for insertion.  They are simply a
+ * pointer to the object after the insertion point.  We store a mark
+ * in the low-bits of the index, to help prevent mistakes.
+ *
+ * When debugging, the index mark changes on every insert and delete, to
+ * catch stale references.
+ */
+#define	INDEX_MAX		(sizeof (uintptr_t) - 1)
+#define	INDEX_NEXT(m)		(((m) == INDEX_MAX)? 1 : ((m) + 1) & INDEX_MAX)
+
+#define	INDEX_TO_NODE(i)	((uu_list_node_impl_t *)((i) & ~INDEX_MAX))
+#define	NODE_TO_INDEX(p, n)	(((uintptr_t)(n) & ~INDEX_MAX) | (p)->ul_index)
+#define	INDEX_VALID(p, i)	(((i) & INDEX_MAX) == (p)->ul_index)
+#define	INDEX_CHECK(i)		(((i) & INDEX_MAX) != 0)
+
+#define	POOL_TO_MARKER(pp) ((void *)((uintptr_t)(pp) | 1))
+
+static uu_list_pool_t	uu_null_lpool = { &uu_null_lpool, &uu_null_lpool };
+static pthread_mutex_t	uu_lpool_list_lock = PTHREAD_MUTEX_INITIALIZER;
+
+uu_list_pool_t *
+uu_list_pool_create(const char *name, size_t objsize,
+    size_t nodeoffset, uu_compare_fn_t *compare_func, uint32_t flags)
+{
+	uu_list_pool_t *pp, *next, *prev;
+
+	if (name == NULL ||
+	    uu_check_name(name, UU_NAME_DOMAIN) == -1 ||
+	    nodeoffset + sizeof (uu_list_node_t) > objsize) {
+		uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+		return (NULL);
+	}
+
+	if (flags & ~UU_LIST_POOL_DEBUG) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	pp = uu_zalloc(sizeof (uu_list_pool_t));
+	if (pp == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	(void) strlcpy(pp->ulp_name, name, sizeof (pp->ulp_name));
+	pp->ulp_nodeoffset = nodeoffset;
+	pp->ulp_objsize = objsize;
+	pp->ulp_cmp = compare_func;
+	if (flags & UU_LIST_POOL_DEBUG)
+		pp->ulp_debug = 1;
+	pp->ulp_last_index = 0;
+
+	(void) pthread_mutex_init(&pp->ulp_lock, NULL);
+
+	pp->ulp_null_list.ul_next_enc = UU_PTR_ENCODE(&pp->ulp_null_list);
+	pp->ulp_null_list.ul_prev_enc = UU_PTR_ENCODE(&pp->ulp_null_list);
+
+	(void) pthread_mutex_lock(&uu_lpool_list_lock);
+	pp->ulp_next = next = &uu_null_lpool;
+	pp->ulp_prev = prev = next->ulp_prev;
+	next->ulp_prev = pp;
+	prev->ulp_next = pp;
+	(void) pthread_mutex_unlock(&uu_lpool_list_lock);
+
+	return (pp);
+}
+
+void
+uu_list_pool_destroy(uu_list_pool_t *pp)
+{
+	if (pp->ulp_debug) {
+		if (pp->ulp_null_list.ul_next_enc !=
+		    UU_PTR_ENCODE(&pp->ulp_null_list) ||
+		    pp->ulp_null_list.ul_prev_enc !=
+		    UU_PTR_ENCODE(&pp->ulp_null_list)) {
+			uu_panic("uu_list_pool_destroy: Pool \"%.*s\" (%p) has "
+			    "outstanding lists, or is corrupt.\n",
+			    (int)sizeof (pp->ulp_name), pp->ulp_name,
+			    (void *)pp);
+		}
+	}
+	(void) pthread_mutex_lock(&uu_lpool_list_lock);
+	pp->ulp_next->ulp_prev = pp->ulp_prev;
+	pp->ulp_prev->ulp_next = pp->ulp_next;
+	(void) pthread_mutex_unlock(&uu_lpool_list_lock);
+	pp->ulp_prev = NULL;
+	pp->ulp_next = NULL;
+	uu_free(pp);
+}
+
+void
+uu_list_node_init(void *base, uu_list_node_t *np_arg, uu_list_pool_t *pp)
+{
+	uu_list_node_impl_t *np = (uu_list_node_impl_t *)np_arg;
+
+	if (pp->ulp_debug) {
+		uintptr_t offset = (uintptr_t)np - (uintptr_t)base;
+		if (offset + sizeof (*np) > pp->ulp_objsize) {
+			uu_panic("uu_list_node_init(%p, %p, %p (\"%s\")): "
+			    "offset %ld doesn't fit in object (size %ld)\n",
+			    base, (void *)np, (void *)pp, pp->ulp_name,
+			    (long)offset, (long)pp->ulp_objsize);
+		}
+		if (offset != pp->ulp_nodeoffset) {
+			uu_panic("uu_list_node_init(%p, %p, %p (\"%s\")): "
+			    "offset %ld doesn't match pool's offset (%ld)\n",
+			    base, (void *)np, (void *)pp, pp->ulp_name,
+			    (long)offset, (long)pp->ulp_objsize);
+		}
+	}
+	np->uln_next = POOL_TO_MARKER(pp);
+	np->uln_prev = NULL;
+}
+
+void
+uu_list_node_fini(void *base, uu_list_node_t *np_arg, uu_list_pool_t *pp)
+{
+	uu_list_node_impl_t *np = (uu_list_node_impl_t *)np_arg;
+
+	if (pp->ulp_debug) {
+		if (np->uln_next == NULL &&
+		    np->uln_prev == NULL) {
+			uu_panic("uu_list_node_fini(%p, %p, %p (\"%s\")): "
+			    "node already finied\n",
+			    base, (void *)np_arg, (void *)pp, pp->ulp_name);
+		}
+		if (np->uln_next != POOL_TO_MARKER(pp) ||
+		    np->uln_prev != NULL) {
+			uu_panic("uu_list_node_fini(%p, %p, %p (\"%s\")): "
+			    "node corrupt or on list\n",
+			    base, (void *)np_arg, (void *)pp, pp->ulp_name);
+		}
+	}
+	np->uln_next = NULL;
+	np->uln_prev = NULL;
+}
+
+uu_list_t *
+uu_list_create(uu_list_pool_t *pp, void *parent, uint32_t flags)
+{
+	uu_list_t *lp, *next, *prev;
+
+	if (flags & ~(UU_LIST_DEBUG | UU_LIST_SORTED)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	if ((flags & UU_LIST_SORTED) && pp->ulp_cmp == NULL) {
+		if (pp->ulp_debug)
+			uu_panic("uu_list_create(%p, ...): requested "
+			    "UU_LIST_SORTED, but pool has no comparison func\n",
+			    (void *)pp);
+		uu_set_error(UU_ERROR_NOT_SUPPORTED);
+		return (NULL);
+	}
+
+	lp = uu_zalloc(sizeof (*lp));
+	if (lp == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	lp->ul_pool = pp;
+	lp->ul_parent_enc = UU_PTR_ENCODE(parent);
+	lp->ul_offset = pp->ulp_nodeoffset;
+	lp->ul_debug = pp->ulp_debug || (flags & UU_LIST_DEBUG);
+	lp->ul_sorted = (flags & UU_LIST_SORTED);
+	lp->ul_numnodes = 0;
+	lp->ul_index = (pp->ulp_last_index = INDEX_NEXT(pp->ulp_last_index));
+
+	lp->ul_null_node.uln_next = &lp->ul_null_node;
+	lp->ul_null_node.uln_prev = &lp->ul_null_node;
+
+	lp->ul_null_walk.ulw_next = &lp->ul_null_walk;
+	lp->ul_null_walk.ulw_prev = &lp->ul_null_walk;
+
+	(void) pthread_mutex_lock(&pp->ulp_lock);
+	next = &pp->ulp_null_list;
+	prev = UU_PTR_DECODE(next->ul_prev_enc);
+	lp->ul_next_enc = UU_PTR_ENCODE(next);
+	lp->ul_prev_enc = UU_PTR_ENCODE(prev);
+	next->ul_prev_enc = UU_PTR_ENCODE(lp);
+	prev->ul_next_enc = UU_PTR_ENCODE(lp);
+	(void) pthread_mutex_unlock(&pp->ulp_lock);
+
+	return (lp);
+}
+
+void
+uu_list_destroy(uu_list_t *lp)
+{
+	uu_list_pool_t *pp = lp->ul_pool;
+
+	if (lp->ul_debug) {
+		if (lp->ul_null_node.uln_next != &lp->ul_null_node ||
+		    lp->ul_null_node.uln_prev != &lp->ul_null_node) {
+			uu_panic("uu_list_destroy(%p):  list not empty\n",
+			    (void *)lp);
+		}
+		if (lp->ul_numnodes != 0) {
+			uu_panic("uu_list_destroy(%p):  numnodes is nonzero, "
+			    "but list is empty\n", (void *)lp);
+		}
+		if (lp->ul_null_walk.ulw_next != &lp->ul_null_walk ||
+		    lp->ul_null_walk.ulw_prev != &lp->ul_null_walk) {
+			uu_panic("uu_list_destroy(%p):  outstanding walkers\n",
+			    (void *)lp);
+		}
+	}
+
+	(void) pthread_mutex_lock(&pp->ulp_lock);
+	UU_LIST_PTR(lp->ul_next_enc)->ul_prev_enc = lp->ul_prev_enc;
+	UU_LIST_PTR(lp->ul_prev_enc)->ul_next_enc = lp->ul_next_enc;
+	(void) pthread_mutex_unlock(&pp->ulp_lock);
+	lp->ul_prev_enc = UU_PTR_ENCODE(NULL);
+	lp->ul_next_enc = UU_PTR_ENCODE(NULL);
+	lp->ul_pool = NULL;
+	uu_free(lp);
+}
+
+static void
+list_insert(uu_list_t *lp, uu_list_node_impl_t *np, uu_list_node_impl_t *prev,
+    uu_list_node_impl_t *next)
+{
+	if (lp->ul_debug) {
+		if (next->uln_prev != prev || prev->uln_next != next)
+			uu_panic("insert(%p): internal error: %p and %p not "
+			    "neighbors\n", (void *)lp, (void *)next,
+			    (void *)prev);
+
+		if (np->uln_next != POOL_TO_MARKER(lp->ul_pool) ||
+		    np->uln_prev != NULL) {
+			uu_panic("insert(%p): elem %p node %p corrupt, "
+			    "not initialized, or already in a list.\n",
+			    (void *)lp, NODE_TO_ELEM(lp, np), (void *)np);
+		}
+		/*
+		 * invalidate outstanding uu_list_index_ts.
+		 */
+		lp->ul_index = INDEX_NEXT(lp->ul_index);
+	}
+	np->uln_next = next;
+	np->uln_prev = prev;
+	next->uln_prev = np;
+	prev->uln_next = np;
+
+	lp->ul_numnodes++;
+}
+
+void
+uu_list_insert(uu_list_t *lp, void *elem, uu_list_index_t idx)
+{
+	uu_list_node_impl_t *np;
+
+	np = INDEX_TO_NODE(idx);
+	if (np == NULL)
+		np = &lp->ul_null_node;
+
+	if (lp->ul_debug) {
+		if (!INDEX_VALID(lp, idx))
+			uu_panic("uu_list_insert(%p, %p, %p): %s\n",
+			    (void *)lp, elem, (void *)idx,
+			    INDEX_CHECK(idx)? "outdated index" :
+			    "invalid index");
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_insert(%p, %p, %p): out-of-date "
+			    "index\n", (void *)lp, elem, (void *)idx);
+	}
+
+	list_insert(lp, ELEM_TO_NODE(lp, elem), np->uln_prev, np);
+}
+
+void *
+uu_list_find(uu_list_t *lp, void *elem, void *private, uu_list_index_t *out)
+{
+	int sorted = lp->ul_sorted;
+	uu_compare_fn_t *func = lp->ul_pool->ulp_cmp;
+	uu_list_node_impl_t *np;
+
+	uu_set_error(UU_ERROR_NONE);
+
+	if (func == NULL) {
+		if (out != NULL)
+			*out = 0;
+		uu_set_error(UU_ERROR_NOT_SUPPORTED);
+		return (NULL);
+	}
+	for (np = lp->ul_null_node.uln_next; np != &lp->ul_null_node;
+	    np = np->uln_next) {
+		void *ep = NODE_TO_ELEM(lp, np);
+		int cmp = func(ep, elem, private);
+		if (cmp == 0) {
+			if (out != NULL)
+				*out = NODE_TO_INDEX(lp, np);
+			return (ep);
+		}
+		if (sorted && cmp > 0) {
+			if (out != NULL)
+				*out = NODE_TO_INDEX(lp, np);
+			return (NULL);
+		}
+	}
+	if (out != NULL)
+		*out = NODE_TO_INDEX(lp, 0);
+	return (NULL);
+}
+
+void *
+uu_list_nearest_next(uu_list_t *lp, uu_list_index_t idx)
+{
+	uu_list_node_impl_t *np = INDEX_TO_NODE(idx);
+
+	if (np == NULL)
+		np = &lp->ul_null_node;
+
+	if (lp->ul_debug) {
+		if (!INDEX_VALID(lp, idx))
+			uu_panic("uu_list_nearest_next(%p, %p): %s\n",
+			    (void *)lp, (void *)idx,
+			    INDEX_CHECK(idx)? "outdated index" :
+			    "invalid index");
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_nearest_next(%p, %p): out-of-date "
+			    "index\n", (void *)lp, (void *)idx);
+	}
+
+	if (np == &lp->ul_null_node)
+		return (NULL);
+	else
+		return (NODE_TO_ELEM(lp, np));
+}
+
+void *
+uu_list_nearest_prev(uu_list_t *lp, uu_list_index_t idx)
+{
+	uu_list_node_impl_t *np = INDEX_TO_NODE(idx);
+
+	if (np == NULL)
+		np = &lp->ul_null_node;
+
+	if (lp->ul_debug) {
+		if (!INDEX_VALID(lp, idx))
+			uu_panic("uu_list_nearest_prev(%p, %p): %s\n",
+			    (void *)lp, (void *)idx, INDEX_CHECK(idx)?
+			    "outdated index" : "invalid index");
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_nearest_prev(%p, %p): out-of-date "
+			    "index\n", (void *)lp, (void *)idx);
+	}
+
+	if ((np = np->uln_prev) == &lp->ul_null_node)
+		return (NULL);
+	else
+		return (NODE_TO_ELEM(lp, np));
+}
+
+static void
+list_walk_init(uu_list_walk_t *wp, uu_list_t *lp, uint32_t flags)
+{
+	uu_list_walk_t *next, *prev;
+
+	int robust = (flags & UU_WALK_ROBUST);
+	int direction = (flags & UU_WALK_REVERSE)? -1 : 1;
+
+	(void) memset(wp, 0, sizeof (*wp));
+	wp->ulw_list = lp;
+	wp->ulw_robust = robust;
+	wp->ulw_dir = direction;
+	if (direction > 0)
+		wp->ulw_next_result = lp->ul_null_node.uln_next;
+	else
+		wp->ulw_next_result = lp->ul_null_node.uln_prev;
+
+	if (lp->ul_debug || robust) {
+		/*
+		 * Add this walker to the list's list of walkers so
+		 * uu_list_remove() can advance us if somebody tries to
+		 * remove ulw_next_result.
+		 */
+		wp->ulw_next = next = &lp->ul_null_walk;
+		wp->ulw_prev = prev = next->ulw_prev;
+		next->ulw_prev = wp;
+		prev->ulw_next = wp;
+	}
+}
+
+static uu_list_node_impl_t *
+list_walk_advance(uu_list_walk_t *wp, uu_list_t *lp)
+{
+	uu_list_node_impl_t *np = wp->ulw_next_result;
+	uu_list_node_impl_t *next;
+
+	if (np == &lp->ul_null_node)
+		return (NULL);
+
+	next = (wp->ulw_dir > 0)? np->uln_next : np->uln_prev;
+
+	wp->ulw_next_result = next;
+	return (np);
+}
+
+static void
+list_walk_fini(uu_list_walk_t *wp)
+{
+	/* GLXXX debugging? */
+	if (wp->ulw_next != NULL) {
+		wp->ulw_next->ulw_prev = wp->ulw_prev;
+		wp->ulw_prev->ulw_next = wp->ulw_next;
+		wp->ulw_next = NULL;
+		wp->ulw_prev = NULL;
+	}
+	wp->ulw_list = NULL;
+	wp->ulw_next_result = NULL;
+}
+
+uu_list_walk_t *
+uu_list_walk_start(uu_list_t *lp, uint32_t flags)
+{
+	uu_list_walk_t *wp;
+
+	if (flags & ~(UU_WALK_ROBUST | UU_WALK_REVERSE)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (NULL);
+	}
+
+	wp = uu_zalloc(sizeof (*wp));
+	if (wp == NULL) {
+		uu_set_error(UU_ERROR_NO_MEMORY);
+		return (NULL);
+	}
+
+	list_walk_init(wp, lp, flags);
+	return (wp);
+}
+
+void *
+uu_list_walk_next(uu_list_walk_t *wp)
+{
+	uu_list_t *lp = wp->ulw_list;
+	uu_list_node_impl_t *np = list_walk_advance(wp, lp);
+
+	if (np == NULL)
+		return (NULL);
+
+	return (NODE_TO_ELEM(lp, np));
+}
+
+void
+uu_list_walk_end(uu_list_walk_t *wp)
+{
+	list_walk_fini(wp);
+	uu_free(wp);
+}
+
+int
+uu_list_walk(uu_list_t *lp, uu_walk_fn_t *func, void *private, uint32_t flags)
+{
+	uu_list_node_impl_t *np;
+
+	int status = UU_WALK_NEXT;
+
+	int robust = (flags & UU_WALK_ROBUST);
+	int reverse = (flags & UU_WALK_REVERSE);
+
+	if (flags & ~(UU_WALK_ROBUST | UU_WALK_REVERSE)) {
+		uu_set_error(UU_ERROR_UNKNOWN_FLAG);
+		return (-1);
+	}
+
+	if (lp->ul_debug || robust) {
+		uu_list_walk_t my_walk;
+		void *e;
+
+		list_walk_init(&my_walk, lp, flags);
+		while (status == UU_WALK_NEXT &&
+		    (e = uu_list_walk_next(&my_walk)) != NULL)
+			status = (*func)(e, private);
+		list_walk_fini(&my_walk);
+	} else {
+		if (!reverse) {
+			for (np = lp->ul_null_node.uln_next;
+			    status == UU_WALK_NEXT && np != &lp->ul_null_node;
+			    np = np->uln_next) {
+				status = (*func)(NODE_TO_ELEM(lp, np), private);
+			}
+		} else {
+			for (np = lp->ul_null_node.uln_prev;
+			    status == UU_WALK_NEXT && np != &lp->ul_null_node;
+			    np = np->uln_prev) {
+				status = (*func)(NODE_TO_ELEM(lp, np), private);
+			}
+		}
+	}
+	if (status >= 0)
+		return (0);
+	uu_set_error(UU_ERROR_CALLBACK_FAILED);
+	return (-1);
+}
+
+void
+uu_list_remove(uu_list_t *lp, void *elem)
+{
+	uu_list_node_impl_t *np = ELEM_TO_NODE(lp, elem);
+	uu_list_walk_t *wp;
+
+	if (lp->ul_debug) {
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_remove(%p, %p): elem not on list\n",
+			    (void *)lp, elem);
+		/*
+		 * invalidate outstanding uu_list_index_ts.
+		 */
+		lp->ul_index = INDEX_NEXT(lp->ul_index);
+	}
+
+	/*
+	 * robust walkers must be advanced.  In debug mode, non-robust
+	 * walkers are also on the list.  If there are any, it's an error.
+	 */
+	for (wp = lp->ul_null_walk.ulw_next; wp != &lp->ul_null_walk;
+	    wp = wp->ulw_next) {
+		if (wp->ulw_robust) {
+			if (np == wp->ulw_next_result)
+				(void) list_walk_advance(wp, lp);
+		} else if (wp->ulw_next_result != NULL) {
+			uu_panic("uu_list_remove(%p, %p): active non-robust "
+			    "walker\n", (void *)lp, elem);
+		}
+	}
+
+	np->uln_next->uln_prev = np->uln_prev;
+	np->uln_prev->uln_next = np->uln_next;
+
+	lp->ul_numnodes--;
+
+	np->uln_next = POOL_TO_MARKER(lp->ul_pool);
+	np->uln_prev = NULL;
+}
+
+void *
+uu_list_teardown(uu_list_t *lp, void **cookie)
+{
+	void *ep;
+
+	/*
+	 * XXX: disable list modification until list is empty
+	 */
+	if (lp->ul_debug && *cookie != NULL)
+		uu_panic("uu_list_teardown(%p, %p): unexpected cookie\n",
+		    (void *)lp, (void *)cookie);
+
+	ep = uu_list_first(lp);
+	if (ep)
+		uu_list_remove(lp, ep);
+	return (ep);
+}
+
+int
+uu_list_insert_before(uu_list_t *lp, void *target, void *elem)
+{
+	uu_list_node_impl_t *np = ELEM_TO_NODE(lp, target);
+
+	if (target == NULL)
+		np = &lp->ul_null_node;
+
+	if (lp->ul_debug) {
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_insert_before(%p, %p, %p): %p is "
+			    "not currently on a list\n",
+			    (void *)lp, target, elem, target);
+	}
+	if (lp->ul_sorted) {
+		if (lp->ul_debug)
+			uu_panic("uu_list_insert_before(%p, ...): list is "
+			    "UU_LIST_SORTED\n", (void *)lp);
+		uu_set_error(UU_ERROR_NOT_SUPPORTED);
+		return (-1);
+	}
+
+	list_insert(lp, ELEM_TO_NODE(lp, elem), np->uln_prev, np);
+	return (0);
+}
+
+int
+uu_list_insert_after(uu_list_t *lp, void *target, void *elem)
+{
+	uu_list_node_impl_t *np = ELEM_TO_NODE(lp, target);
+
+	if (target == NULL)
+		np = &lp->ul_null_node;
+
+	if (lp->ul_debug) {
+		if (np->uln_prev == NULL)
+			uu_panic("uu_list_insert_after(%p, %p, %p): %p is "
+			    "not currently on a list\n",
+			    (void *)lp, target, elem, target);
+	}
+	if (lp->ul_sorted) {
+		if (lp->ul_debug)
+			uu_panic("uu_list_insert_after(%p, ...): list is "
+			    "UU_LIST_SORTED\n", (void *)lp);
+		uu_set_error(UU_ERROR_NOT_SUPPORTED);
+		return (-1);
+	}
+
+	list_insert(lp, ELEM_TO_NODE(lp, elem), np, np->uln_next);
+	return (0);
+}
+
+size_t
+uu_list_numnodes(uu_list_t *lp)
+{
+	return (lp->ul_numnodes);
+}
+
+void *
+uu_list_first(uu_list_t *lp)
+{
+	uu_list_node_impl_t *n = lp->ul_null_node.uln_next;
+	if (n == &lp->ul_null_node)
+		return (NULL);
+	return (NODE_TO_ELEM(lp, n));
+}
+
+void *
+uu_list_last(uu_list_t *lp)
+{
+	uu_list_node_impl_t *n = lp->ul_null_node.uln_prev;
+	if (n == &lp->ul_null_node)
+		return (NULL);
+	return (NODE_TO_ELEM(lp, n));
+}
+
+void *
+uu_list_next(uu_list_t *lp, void *elem)
+{
+	uu_list_node_impl_t *n = ELEM_TO_NODE(lp, elem);
+
+	n = n->uln_next;
+	if (n == &lp->ul_null_node)
+		return (NULL);
+	return (NODE_TO_ELEM(lp, n));
+}
+
+void *
+uu_list_prev(uu_list_t *lp, void *elem)
+{
+	uu_list_node_impl_t *n = ELEM_TO_NODE(lp, elem);
+
+	n = n->uln_prev;
+	if (n == &lp->ul_null_node)
+		return (NULL);
+	return (NODE_TO_ELEM(lp, n));
+}
+
+/*
+ * called from uu_lockup() and uu_release(), as part of our fork1()-safety.
+ */
+void
+uu_list_lockup(void)
+{
+	uu_list_pool_t *pp;
+
+	(void) pthread_mutex_lock(&uu_lpool_list_lock);
+	for (pp = uu_null_lpool.ulp_next; pp != &uu_null_lpool;
+	    pp = pp->ulp_next)
+		(void) pthread_mutex_lock(&pp->ulp_lock);
+}
+
+void
+uu_list_release(void)
+{
+	uu_list_pool_t *pp;
+
+	for (pp = uu_null_lpool.ulp_next; pp != &uu_null_lpool;
+	    pp = pp->ulp_next)
+		(void) pthread_mutex_unlock(&pp->ulp_lock);
+	(void) pthread_mutex_unlock(&uu_lpool_list_lock);
+}

lib/libuutil/common/uu_misc.c

@@ -0,0 +1,280 @@
+/*
+ * 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 (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#include "libuutil_common.h"
+
+#include <assert.h>
+#include <errno.h>
+#include <libintl.h>
+#include <pthread.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/debug.h>
+#include <thread.h>
+#include <unistd.h>
+#include <ctype.h>
+
+#if !defined(TEXT_DOMAIN)
+#define	TEXT_DOMAIN "SYS_TEST"
+#endif
+
+/*
+ * All of the old code under !defined(PTHREAD_ONCE_KEY_NP)
+ * is here to enable the building of a native version of
+ * libuutil.so when the build machine has not yet been upgraded
+ * to a version of libc that provides pthread_key_create_once_np().
+ * It should all be deleted when solaris_nevada ships.
+ * The code is not MT-safe in a relaxed memory model.
+ */
+
+#if defined(PTHREAD_ONCE_KEY_NP)
+static pthread_key_t	uu_error_key = PTHREAD_ONCE_KEY_NP;
+#else	/* PTHREAD_ONCE_KEY_NP */
+static pthread_key_t	uu_error_key = 0;
+static pthread_mutex_t	uu_key_lock = PTHREAD_MUTEX_INITIALIZER;
+#endif	/* PTHREAD_ONCE_KEY_NP */
+
+static int		uu_error_key_setup = 0;
+
+static pthread_mutex_t	uu_panic_lock = PTHREAD_MUTEX_INITIALIZER;
+/* LINTED static unused */
+static const char	*uu_panic_format;
+/* LINTED static unused */
+static va_list		uu_panic_args;
+static pthread_t	uu_panic_thread;
+
+static uint32_t		_uu_main_error;
+
+void
+uu_set_error(uint_t code)
+{
+	if (thr_main() != 0) {
+		_uu_main_error = code;
+		return;
+	}
+#if defined(PTHREAD_ONCE_KEY_NP)
+	if (pthread_key_create_once_np(&uu_error_key, NULL) != 0)
+		uu_error_key_setup = -1;
+	else
+		uu_error_key_setup = 1;
+#else	/* PTHREAD_ONCE_KEY_NP */
+	if (uu_error_key_setup == 0) {
+		(void) pthread_mutex_lock(&uu_key_lock);
+		if (uu_error_key_setup == 0) {
+			if (pthread_key_create(&uu_error_key, NULL) != 0)
+				uu_error_key_setup = -1;
+			else
+				uu_error_key_setup = 1;
+		}
+		(void) pthread_mutex_unlock(&uu_key_lock);
+	}
+#endif	/* PTHREAD_ONCE_KEY_NP */
+	if (uu_error_key_setup > 0)
+		(void) pthread_setspecific(uu_error_key,
+		    (void *)(uintptr_t)code);
+}
+
+uint32_t
+uu_error(void)
+{
+	if (thr_main() != 0)
+		return (_uu_main_error);
+
+	if (uu_error_key_setup < 0)	/* can't happen? */
+		return (UU_ERROR_UNKNOWN);
+
+	/*
+	 * Because UU_ERROR_NONE == 0, if uu_set_error() was
+	 * never called, then this will return UU_ERROR_NONE:
+	 */
+	return ((uint32_t)(uintptr_t)pthread_getspecific(uu_error_key));
+}
+
+const char *
+uu_strerror(uint32_t code)
+{
+	const char *str;
+
+	switch (code) {
+	case UU_ERROR_NONE:
+		str = dgettext(TEXT_DOMAIN, "No error");
+		break;
+
+	case UU_ERROR_INVALID_ARGUMENT:
+		str = dgettext(TEXT_DOMAIN, "Invalid argument");
+		break;
+
+	case UU_ERROR_UNKNOWN_FLAG:
+		str = dgettext(TEXT_DOMAIN, "Unknown flag passed");
+		break;
+
+	case UU_ERROR_NO_MEMORY:
+		str = dgettext(TEXT_DOMAIN, "Out of memory");
+		break;
+
+	case UU_ERROR_CALLBACK_FAILED:
+		str = dgettext(TEXT_DOMAIN, "Callback-initiated failure");
+		break;
+
+	case UU_ERROR_NOT_SUPPORTED:
+		str = dgettext(TEXT_DOMAIN, "Operation not supported");
+		break;
+
+	case UU_ERROR_EMPTY:
+		str = dgettext(TEXT_DOMAIN, "No value provided");
+		break;
+
+	case UU_ERROR_UNDERFLOW:
+		str = dgettext(TEXT_DOMAIN, "Value too small");
+		break;
+
+	case UU_ERROR_OVERFLOW:
+		str = dgettext(TEXT_DOMAIN, "Value too large");
+		break;
+
+	case UU_ERROR_INVALID_CHAR:
+		str = dgettext(TEXT_DOMAIN,
+		    "Value contains unexpected character");
+		break;
+
+	case UU_ERROR_INVALID_DIGIT:
+		str = dgettext(TEXT_DOMAIN,
+		    "Value contains digit not in base");
+		break;
+
+	case UU_ERROR_SYSTEM:
+		str = dgettext(TEXT_DOMAIN, "Underlying system error");
+		break;
+
+	case UU_ERROR_UNKNOWN:
+		str = dgettext(TEXT_DOMAIN, "Error status not known");
+		break;
+
+	default:
+		errno = ESRCH;
+		str = NULL;
+		break;
+	}
+	return (str);
+}
+
+void
+uu_panic(const char *format, ...)
+{
+	va_list args;
+
+	va_start(args, format);
+
+	(void) pthread_mutex_lock(&uu_panic_lock);
+	if (uu_panic_thread == 0) {
+		uu_panic_thread = pthread_self();
+		uu_panic_format = format;
+		va_copy(uu_panic_args, args);
+	}
+	(void) pthread_mutex_unlock(&uu_panic_lock);
+
+	(void) vfprintf(stderr, format, args);
+
+	if (uu_panic_thread == pthread_self())
+		abort();
+	else
+		for (;;)
+			(void) pause();
+}
+
+int
+assfail(const char *astring, const char *file, int line)
+{
+	__assert(astring, file, line);
+	/*NOTREACHED*/
+	return (0);
+}
+
+static void
+uu_lockup(void)
+{
+	(void) pthread_mutex_lock(&uu_panic_lock);
+#if !defined(PTHREAD_ONCE_KEY_NP)
+	(void) pthread_mutex_lock(&uu_key_lock);
+#endif
+	uu_avl_lockup();
+	uu_list_lockup();
+}
+
+static void
+uu_release(void)
+{
+	(void) pthread_mutex_unlock(&uu_panic_lock);
+#if !defined(PTHREAD_ONCE_KEY_NP)
+	(void) pthread_mutex_unlock(&uu_key_lock);
+#endif
+	uu_avl_release();
+	uu_list_release();
+}
+
+static void
+uu_release_child(void)
+{
+	uu_panic_format = NULL;
+	uu_panic_thread = 0;
+
+	uu_release();
+}
+
+#pragma init(uu_init)
+static void
+uu_init(void)
+{
+	(void) pthread_atfork(uu_lockup, uu_release, uu_release_child);
+}
+
+/*
+ * Dump a block of memory in hex+ascii, for debugging
+ */
+void
+uu_dump(FILE *out, const char *prefix, const void *buf, size_t len)
+{
+	const unsigned char *p = buf;
+	int i;
+
+	for (i = 0; i < len; i += 16) {
+		int j;
+
+		(void) fprintf(out, "%s", prefix);
+		for (j = 0; j < 16 && i + j < len; j++) {
+			(void) fprintf(out, "%2.2x ", p[i + j]);
+		}
+		for (; j < 16; j++) {
+			(void) fprintf(out, "   ");
+		}
+		for (j = 0; j < 16 && i + j < len; j++) {
+			(void) fprintf(out, "%c",
+			    isprint(p[i + j]) ? p[i + j] : '.');
+		}
+		(void) fprintf(out, "\n");
+	}
+}

lib/libuutil/common/uu_open.c

@@ -0,0 +1,70 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2004 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <sys/time.h>
+
+#include <errno.h>
+#include <fcntl.h>
+#include <limits.h>
+#include <stdio.h>
+#include <unistd.h>
+
+#ifdef _LP64
+#define	TMPPATHFMT	"%s/uu%ld"
+#else /* _LP64 */
+#define	TMPPATHFMT	"%s/uu%lld"
+#endif /* _LP64 */
+
+/*ARGSUSED*/
+int
+uu_open_tmp(const char *dir, uint_t uflags)
+{
+	int f;
+	char *fname = uu_zalloc(PATH_MAX);
+
+	if (fname == NULL)
+		return (-1);
+
+	for (;;) {
+		(void) snprintf(fname, PATH_MAX, "%s/uu%lld", dir, gethrtime());
+
+		f = open(fname, O_CREAT | O_EXCL | O_RDWR, 0600);
+
+		if (f >= 0 || errno != EEXIST)
+			break;
+	}
+
+	if (f >= 0)
+		(void) unlink(fname);
+
+	uu_free(fname);
+
+	return (f);
+}

lib/libuutil/common/uu_pname.c

@@ -0,0 +1,207 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2005 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <libintl.h>
+#include <limits.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <errno.h>
+#include <wchar.h>
+#include <unistd.h>
+
+static const char PNAME_FMT[] = "%s: ";
+static const char ERRNO_FMT[] = ": %s\n";
+
+static const char *pname;
+
+static void
+uu_die_internal(int status, const char *format, va_list alist) __NORETURN;
+
+int uu_exit_ok_value = EXIT_SUCCESS;
+int uu_exit_fatal_value = EXIT_FAILURE;
+int uu_exit_usage_value = 2;
+
+int *
+uu_exit_ok(void)
+{
+	return (&uu_exit_ok_value);
+}
+
+int *
+uu_exit_fatal(void)
+{
+	return (&uu_exit_fatal_value);
+}
+
+int *
+uu_exit_usage(void)
+{
+	return (&uu_exit_usage_value);
+}
+
+void
+uu_alt_exit(int profile)
+{
+	switch (profile) {
+	case UU_PROFILE_DEFAULT:
+		uu_exit_ok_value = EXIT_SUCCESS;
+		uu_exit_fatal_value = EXIT_FAILURE;
+		uu_exit_usage_value = 2;
+		break;
+	case UU_PROFILE_LAUNCHER:
+		uu_exit_ok_value = EXIT_SUCCESS;
+		uu_exit_fatal_value = 124;
+		uu_exit_usage_value = 125;
+		break;
+	}
+}
+
+static void
+uu_warn_internal(int err, const char *format, va_list alist)
+{
+	if (pname != NULL)
+		(void) fprintf(stderr, PNAME_FMT, pname);
+
+	(void) vfprintf(stderr, format, alist);
+
+	if (strrchr(format, '\n') == NULL)
+		(void) fprintf(stderr, ERRNO_FMT, strerror(err));
+}
+
+void
+uu_vwarn(const char *format, va_list alist)
+{
+	uu_warn_internal(errno, format, alist);
+}
+
+/*PRINTFLIKE1*/
+void
+uu_warn(const char *format, ...)
+{
+	va_list alist;
+	va_start(alist, format);
+	uu_warn_internal(errno, format, alist);
+	va_end(alist);
+}
+
+static void
+uu_die_internal(int status, const char *format, va_list alist)
+{
+	uu_warn_internal(errno, format, alist);
+#ifdef DEBUG
+	{
+		char *cp;
+
+		if (!issetugid()) {
+			cp = getenv("UU_DIE_ABORTS");
+			if (cp != NULL && *cp != '\0')
+				abort();
+		}
+	}
+#endif
+	exit(status);
+}
+
+void
+uu_vdie(const char *format, va_list alist)
+{
+	uu_die_internal(UU_EXIT_FATAL, format, alist);
+}
+
+/*PRINTFLIKE1*/
+void
+uu_die(const char *format, ...)
+{
+	va_list alist;
+	va_start(alist, format);
+	uu_die_internal(UU_EXIT_FATAL, format, alist);
+	va_end(alist);
+}
+
+void
+uu_vxdie(int status, const char *format, va_list alist)
+{
+	uu_die_internal(status, format, alist);
+}
+
+/*PRINTFLIKE2*/
+void
+uu_xdie(int status, const char *format, ...)
+{
+	va_list alist;
+	va_start(alist, format);
+	uu_die_internal(status, format, alist);
+	va_end(alist);
+}
+
+const char *
+uu_setpname(char *arg0)
+{
+	/*
+	 * Having a NULL argv[0], while uncommon, is possible.  It
+	 * makes more sense to handle this event in uu_setpname rather
+	 * than in each of its consumers.
+	 */
+	if (arg0 == NULL) {
+		pname = getexecname();
+		if (pname == NULL)
+			pname = "unknown_command";
+		return (pname);
+	}
+
+	/*
+	 * Guard against '/' at end of command invocation.
+	 */
+	for (;;) {
+		char *p = strrchr(arg0, '/');
+		if (p == NULL) {
+			pname = arg0;
+			break;
+		} else {
+			if (*(p + 1) == '\0') {
+				*p = '\0';
+				continue;
+			}
+
+			pname = p + 1;
+			break;
+		}
+	}
+
+	return (pname);
+}
+
+const char *
+uu_getpname(void)
+{
+	return (pname);
+}

lib/libuutil/common/uu_string.c

@@ -0,0 +1,56 @@
+/*
+ * 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+/*
+ * String helper functions
+ */
+
+#include <string.h>
+#include <sys/types.h>
+#include <stdio.h>
+#include <malloc.h>
+#include <ctype.h>
+#include "libuutil.h"
+
+/* Return true if strings are equal */
+boolean_t
+uu_streq(const char *a, const char *b)
+{
+	return (strcmp(a, b) == 0);
+}
+
+/* Return true if strings are equal, case-insensitively */
+boolean_t
+uu_strcaseeq(const char *a, const char *b)
+{
+	return (strcasecmp(a, b) == 0);
+}
+
+/* Return true if string a Begins With string b */
+boolean_t
+uu_strbw(const char *a, const char *b)
+{
+	return (strncmp(a, b, strlen(b)) == 0);
+}

lib/libuutil/common/uu_strtoint.c

@@ -0,0 +1,300 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (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 2004 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#pragma ident	"%Z%%M%	%I%	%E% SMI"
+
+#include "libuutil_common.h"
+
+#include <limits.h>
+#include <ctype.h>
+
+#define	MAX_BASE	36
+
+#define	IS_DIGIT(x)	((x) >= '0' && (x) <= '9')
+
+#define	CTOI(x) (((x) >= '0' && (x) <= '9') ? (x) - '0' : \
+	    ((x) >= 'a' && (x) <= 'z') ? (x) + 10 - 'a' : (x) + 10 - 'A')
+
+static int
+strtoint(const char *s_arg, uint64_t *out, uint32_t base, int sign)
+{
+	const unsigned char *s = (const unsigned char *)s_arg;
+
+	uint64_t val = 0;
+	uint64_t multmax;
+
+	unsigned c, i;
+
+	int neg = 0;
+
+	int bad_digit = 0;
+	int bad_char = 0;
+	int overflow = 0;
+
+	if (s == NULL || base == 1 || base > MAX_BASE) {
+		uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+		return (-1);
+	}
+
+	while ((c = *s) != 0 && isspace(c))
+		s++;
+
+	switch (c) {
+	case '-':
+		if (!sign)
+			overflow = 1;		/* becomes underflow below */
+		neg = 1;
+		/*FALLTHRU*/
+	case '+':
+		c = *++s;
+		break;
+	default:
+		break;
+	}
+
+	if (c == '\0') {
+		uu_set_error(UU_ERROR_EMPTY);
+		return (-1);
+	}
+
+	if (base == 0) {
+		if (c != '0')
+			base = 10;
+		else if (s[1] == 'x' || s[1] == 'X')
+			base = 16;
+		else
+			base = 8;
+	}
+
+	if (base == 16 && c == '0' && (s[1] == 'x' || s[1] == 'X'))
+		c = *(s += 2);
+
+	if ((val = CTOI(c)) >= base) {
+		if (IS_DIGIT(c))
+			bad_digit = 1;
+		else
+			bad_char = 1;
+		val = 0;
+	}
+
+	multmax = (uint64_t)UINT64_MAX / (uint64_t)base;
+
+	for (c = *++s; c != '\0'; c = *++s) {
+		if ((i = CTOI(c)) >= base) {
+			if (isspace(c))
+				break;
+			if (IS_DIGIT(c))
+				bad_digit = 1;
+			else
+				bad_char = 1;
+			i = 0;
+		}
+
+		if (val > multmax)
+			overflow = 1;
+
+		val *= base;
+		if ((uint64_t)UINT64_MAX - val < (uint64_t)i)
+			overflow = 1;
+
+		val += i;
+	}
+
+	while ((c = *s) != 0) {
+		if (!isspace(c))
+			bad_char = 1;
+		s++;
+	}
+
+	if (sign) {
+		if (neg) {
+			if (val > -(uint64_t)INT64_MIN)
+				overflow = 1;
+		} else {
+			if (val > INT64_MAX)
+				overflow = 1;
+		}
+	}
+
+	if (neg)
+		val = -val;
+
+	if (bad_char | bad_digit | overflow) {
+		if (bad_char)
+			uu_set_error(UU_ERROR_INVALID_CHAR);
+		else if (bad_digit)
+			uu_set_error(UU_ERROR_INVALID_DIGIT);
+		else if (overflow) {
+			if (neg)
+				uu_set_error(UU_ERROR_UNDERFLOW);
+			else
+				uu_set_error(UU_ERROR_OVERFLOW);
+		}
+		return (-1);
+	}
+
+	*out = val;
+	return (0);
+}
+
+int
+uu_strtoint(const char *s, void *v, size_t sz, int base,
+    int64_t min, int64_t max)
+{
+	uint64_t val_u;
+	int64_t val;
+
+	if (min > max)
+		goto bad_argument;
+
+	switch (sz) {
+	case 1:
+		if (max > INT8_MAX || min < INT8_MIN)
+			goto bad_argument;
+		break;
+	case 2:
+		if (max > INT16_MAX || min < INT16_MIN)
+			goto bad_argument;
+		break;
+	case 4:
+		if (max > INT32_MAX || min < INT32_MIN)
+			goto bad_argument;
+		break;
+	case 8:
+		if (max > INT64_MAX || min < INT64_MIN)
+			goto bad_argument;
+		break;
+	default:
+		goto bad_argument;
+	}
+
+	if (min == 0 && max == 0) {
+		min = -(1ULL << (8 * sz - 1));
+		max = (1ULL << (8 * sz - 1)) - 1;
+	}
+
+	if (strtoint(s, &val_u, base, 1) == -1)
+		return (-1);
+
+	val = (int64_t)val_u;
+
+	if (val < min) {
+		uu_set_error(UU_ERROR_UNDERFLOW);
+		return (-1);
+	} else if (val > max) {
+		uu_set_error(UU_ERROR_OVERFLOW);
+		return (-1);
+	}
+
+	switch (sz) {
+	case 1:
+		*(int8_t *)v = val;
+		return (0);
+	case 2:
+		*(int16_t *)v = val;
+		return (0);
+	case 4:
+		*(int32_t *)v = val;
+		return (0);
+	case 8:
+		*(int64_t *)v = val;
+		return (0);
+	default:
+		break;		/* fall through to bad_argument */
+	}
+
+bad_argument:
+	uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+	return (-1);
+}
+
+int
+uu_strtouint(const char *s, void *v, size_t sz, int base,
+    uint64_t min, uint64_t max)
+{
+	uint64_t val;
+
+	if (min > max)
+		goto bad_argument;
+
+	switch (sz) {
+	case 1:
+		if (max > UINT8_MAX)
+			goto bad_argument;
+		break;
+	case 2:
+		if (max > UINT16_MAX)
+			goto bad_argument;
+		break;
+	case 4:
+		if (max > UINT32_MAX)
+			goto bad_argument;
+		break;
+	case 8:
+		if (max > UINT64_MAX)
+			goto bad_argument;
+		break;
+	default:
+		goto bad_argument;
+	}
+
+	if (min == 0 && max == 0) {
+		/* we have to be careful, since << can overflow */
+		max = (1ULL << (8 * sz - 1)) * 2 - 1;
+	}
+
+	if (strtoint(s, &val, base, 0) == -1)
+		return (-1);
+
+	if (val < min) {
+		uu_set_error(UU_ERROR_UNDERFLOW);
+		return (-1);
+	} else if (val > max) {
+		uu_set_error(UU_ERROR_OVERFLOW);
+		return (-1);
+	}
+
+	switch (sz) {
+	case 1:
+		*(uint8_t *)v = val;
+		return (0);
+	case 2:
+		*(uint16_t *)v = val;
+		return (0);
+	case 4:
+		*(uint32_t *)v = val;
+		return (0);
+	case 8:
+		*(uint64_t *)v = val;
+		return (0);
+	default:
+		break;		/* shouldn't happen, fall through */
+	}
+
+bad_argument:
+	uu_set_error(UU_ERROR_INVALID_ARGUMENT);
+	return (-1);
+}

lib/libzfs/common/libzfs.h

@@ -0,0 +1,755 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
+ * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2012, Joyent, Inc. All rights reserved.
+ */
+
+#ifndef	_LIBZFS_H
+#define	_LIBZFS_H
+
+#include <assert.h>
+#include <libnvpair.h>
+#include <sys/mnttab.h>
+#include <sys/param.h>
+#include <sys/types.h>
+#include <sys/varargs.h>
+#include <sys/fs/zfs.h>
+#include <sys/avl.h>
+#include <ucred.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * Miscellaneous ZFS constants
+ */
+#define	ZFS_MAXNAMELEN		MAXNAMELEN
+#define	ZPOOL_MAXNAMELEN	MAXNAMELEN
+#define	ZFS_MAXPROPLEN		MAXPATHLEN
+#define	ZPOOL_MAXPROPLEN	MAXPATHLEN
+
+/*
+ * libzfs errors
+ */
+enum {
+	EZFS_NOMEM = 2000,	/* out of memory */
+	EZFS_BADPROP,		/* invalid property value */
+	EZFS_PROPREADONLY,	/* cannot set readonly property */
+	EZFS_PROPTYPE,		/* property does not apply to dataset type */
+	EZFS_PROPNONINHERIT,	/* property is not inheritable */
+	EZFS_PROPSPACE,		/* bad quota or reservation */
+	EZFS_BADTYPE,		/* dataset is not of appropriate type */
+	EZFS_BUSY,		/* pool or dataset is busy */
+	EZFS_EXISTS,		/* pool or dataset already exists */
+	EZFS_NOENT,		/* no such pool or dataset */
+	EZFS_BADSTREAM,		/* bad backup stream */
+	EZFS_DSREADONLY,	/* dataset is readonly */
+	EZFS_VOLTOOBIG,		/* volume is too large for 32-bit system */
+	EZFS_INVALIDNAME,	/* invalid dataset name */
+	EZFS_BADRESTORE,	/* unable to restore to destination */
+	EZFS_BADBACKUP,		/* backup failed */
+	EZFS_BADTARGET,		/* bad attach/detach/replace target */
+	EZFS_NODEVICE,		/* no such device in pool */
+	EZFS_BADDEV,		/* invalid device to add */
+	EZFS_NOREPLICAS,	/* no valid replicas */
+	EZFS_RESILVERING,	/* currently resilvering */
+	EZFS_BADVERSION,	/* unsupported version */
+	EZFS_POOLUNAVAIL,	/* pool is currently unavailable */
+	EZFS_DEVOVERFLOW,	/* too many devices in one vdev */
+	EZFS_BADPATH,		/* must be an absolute path */
+	EZFS_CROSSTARGET,	/* rename or clone across pool or dataset */
+	EZFS_ZONED,		/* used improperly in local zone */
+	EZFS_MOUNTFAILED,	/* failed to mount dataset */
+	EZFS_UMOUNTFAILED,	/* failed to unmount dataset */
+	EZFS_UNSHARENFSFAILED,	/* unshare(1M) failed */
+	EZFS_SHARENFSFAILED,	/* share(1M) failed */
+	EZFS_PERM,		/* permission denied */
+	EZFS_NOSPC,		/* out of space */
+	EZFS_FAULT,		/* bad address */
+	EZFS_IO,		/* I/O error */
+	EZFS_INTR,		/* signal received */
+	EZFS_ISSPARE,		/* device is a hot spare */
+	EZFS_INVALCONFIG,	/* invalid vdev configuration */
+	EZFS_RECURSIVE,		/* recursive dependency */
+	EZFS_NOHISTORY,		/* no history object */
+	EZFS_POOLPROPS,		/* couldn't retrieve pool props */
+	EZFS_POOL_NOTSUP,	/* ops not supported for this type of pool */
+	EZFS_POOL_INVALARG,	/* invalid argument for this pool operation */
+	EZFS_NAMETOOLONG,	/* dataset name is too long */
+	EZFS_OPENFAILED,	/* open of device failed */
+	EZFS_NOCAP,		/* couldn't get capacity */
+	EZFS_LABELFAILED,	/* write of label failed */
+	EZFS_BADWHO,		/* invalid permission who */
+	EZFS_BADPERM,		/* invalid permission */
+	EZFS_BADPERMSET,	/* invalid permission set name */
+	EZFS_NODELEGATION,	/* delegated administration is disabled */
+	EZFS_UNSHARESMBFAILED,	/* failed to unshare over smb */
+	EZFS_SHARESMBFAILED,	/* failed to share over smb */
+	EZFS_BADCACHE,		/* bad cache file */
+	EZFS_ISL2CACHE,		/* device is for the level 2 ARC */
+	EZFS_VDEVNOTSUP,	/* unsupported vdev type */
+	EZFS_NOTSUP,		/* ops not supported on this dataset */
+	EZFS_ACTIVE_SPARE,	/* pool has active shared spare devices */
+	EZFS_UNPLAYED_LOGS,	/* log device has unplayed logs */
+	EZFS_REFTAG_RELE,	/* snapshot release: tag not found */
+	EZFS_REFTAG_HOLD,	/* snapshot hold: tag already exists */
+	EZFS_TAGTOOLONG,	/* snapshot hold/rele: tag too long */
+	EZFS_PIPEFAILED,	/* pipe create failed */
+	EZFS_THREADCREATEFAILED, /* thread create failed */
+	EZFS_POSTSPLIT_ONLINE,	/* onlining a disk after splitting it */
+	EZFS_SCRUBBING,		/* currently scrubbing */
+	EZFS_NO_SCRUB,		/* no active scrub */
+	EZFS_DIFF,		/* general failure of zfs diff */
+	EZFS_DIFFDATA,		/* bad zfs diff data */
+	EZFS_POOLREADONLY,	/* pool is in read-only mode */
+	EZFS_UNKNOWN
+};
+
+/*
+ * The following data structures are all part
+ * of the zfs_allow_t data structure which is
+ * used for printing 'allow' permissions.
+ * It is a linked list of zfs_allow_t's which
+ * then contain avl tree's for user/group/sets/...
+ * and each one of the entries in those trees have
+ * avl tree's for the permissions they belong to and
+ * whether they are local,descendent or local+descendent
+ * permissions.  The AVL trees are used primarily for
+ * sorting purposes, but also so that we can quickly find
+ * a given user and or permission.
+ */
+typedef struct zfs_perm_node {
+	avl_node_t z_node;
+	char z_pname[MAXPATHLEN];
+} zfs_perm_node_t;
+
+typedef struct zfs_allow_node {
+	avl_node_t z_node;
+	char z_key[MAXPATHLEN];		/* name, such as joe */
+	avl_tree_t z_localdescend;	/* local+descendent perms */
+	avl_tree_t z_local;		/* local permissions */
+	avl_tree_t z_descend;		/* descendent permissions */
+} zfs_allow_node_t;
+
+typedef struct zfs_allow {
+	struct zfs_allow *z_next;
+	char z_setpoint[MAXPATHLEN];
+	avl_tree_t z_sets;
+	avl_tree_t z_crperms;
+	avl_tree_t z_user;
+	avl_tree_t z_group;
+	avl_tree_t z_everyone;
+} zfs_allow_t;
+
+/*
+ * Basic handle types
+ */
+typedef struct zfs_handle zfs_handle_t;
+typedef struct zpool_handle zpool_handle_t;
+typedef struct libzfs_handle libzfs_handle_t;
+
+/*
+ * Library initialization
+ */
+extern libzfs_handle_t *libzfs_init(void);
+extern void libzfs_fini(libzfs_handle_t *);
+
+extern libzfs_handle_t *zpool_get_handle(zpool_handle_t *);
+extern libzfs_handle_t *zfs_get_handle(zfs_handle_t *);
+
+extern void libzfs_print_on_error(libzfs_handle_t *, boolean_t);
+
+extern int libzfs_errno(libzfs_handle_t *);
+extern const char *libzfs_error_action(libzfs_handle_t *);
+extern const char *libzfs_error_description(libzfs_handle_t *);
+extern void libzfs_mnttab_init(libzfs_handle_t *);
+extern void libzfs_mnttab_fini(libzfs_handle_t *);
+extern void libzfs_mnttab_cache(libzfs_handle_t *, boolean_t);
+extern int libzfs_mnttab_find(libzfs_handle_t *, const char *,
+    struct mnttab *);
+extern void libzfs_mnttab_add(libzfs_handle_t *, const char *,
+    const char *, const char *);
+extern void libzfs_mnttab_remove(libzfs_handle_t *, const char *);
+
+/*
+ * Basic handle functions
+ */
+extern zpool_handle_t *zpool_open(libzfs_handle_t *, const char *);
+extern zpool_handle_t *zpool_open_canfail(libzfs_handle_t *, const char *);
+extern void zpool_close(zpool_handle_t *);
+extern const char *zpool_get_name(zpool_handle_t *);
+extern int zpool_get_state(zpool_handle_t *);
+extern char *zpool_state_to_name(vdev_state_t, vdev_aux_t);
+extern void zpool_free_handles(libzfs_handle_t *);
+
+/*
+ * Iterate over all active pools in the system.
+ */
+typedef int (*zpool_iter_f)(zpool_handle_t *, void *);
+extern int zpool_iter(libzfs_handle_t *, zpool_iter_f, void *);
+
+/*
+ * Functions to create and destroy pools
+ */
+extern int zpool_create(libzfs_handle_t *, const char *, nvlist_t *,
+    nvlist_t *, nvlist_t *);
+extern int zpool_destroy(zpool_handle_t *);
+extern int zpool_add(zpool_handle_t *, nvlist_t *);
+
+typedef struct splitflags {
+	/* do not split, but return the config that would be split off */
+	int dryrun : 1;
+
+	/* after splitting, import the pool */
+	int import : 1;
+} splitflags_t;
+
+/*
+ * Functions to manipulate pool and vdev state
+ */
+extern int zpool_scan(zpool_handle_t *, pool_scan_func_t);
+extern int zpool_clear(zpool_handle_t *, const char *, nvlist_t *);
+extern int zpool_reguid(zpool_handle_t *);
+extern int zpool_reopen(zpool_handle_t *);
+
+extern int zpool_vdev_online(zpool_handle_t *, const char *, int,
+    vdev_state_t *);
+extern int zpool_vdev_offline(zpool_handle_t *, const char *, boolean_t);
+extern int zpool_vdev_attach(zpool_handle_t *, const char *,
+    const char *, nvlist_t *, int);
+extern int zpool_vdev_detach(zpool_handle_t *, const char *);
+extern int zpool_vdev_remove(zpool_handle_t *, const char *);
+extern int zpool_vdev_split(zpool_handle_t *, char *, nvlist_t **, nvlist_t *,
+    splitflags_t);
+
+extern int zpool_vdev_fault(zpool_handle_t *, uint64_t, vdev_aux_t);
+extern int zpool_vdev_degrade(zpool_handle_t *, uint64_t, vdev_aux_t);
+extern int zpool_vdev_clear(zpool_handle_t *, uint64_t);
+
+extern nvlist_t *zpool_find_vdev(zpool_handle_t *, const char *, boolean_t *,
+    boolean_t *, boolean_t *);
+extern nvlist_t *zpool_find_vdev_by_physpath(zpool_handle_t *, const char *,
+    boolean_t *, boolean_t *, boolean_t *);
+extern int zpool_label_disk(libzfs_handle_t *, zpool_handle_t *, char *);
+
+/*
+ * Functions to manage pool properties
+ */
+extern int zpool_set_prop(zpool_handle_t *, const char *, const char *);
+extern int zpool_get_prop(zpool_handle_t *, zpool_prop_t, char *,
+    size_t proplen, zprop_source_t *);
+extern uint64_t zpool_get_prop_int(zpool_handle_t *, zpool_prop_t,
+    zprop_source_t *);
+
+extern const char *zpool_prop_to_name(zpool_prop_t);
+extern const char *zpool_prop_values(zpool_prop_t);
+
+/*
+ * Pool health statistics.
+ */
+typedef enum {
+	/*
+	 * The following correspond to faults as defined in the (fault.fs.zfs.*)
+	 * event namespace.  Each is associated with a corresponding message ID.
+	 */
+	ZPOOL_STATUS_CORRUPT_CACHE,	/* corrupt /kernel/drv/zpool.cache */
+	ZPOOL_STATUS_MISSING_DEV_R,	/* missing device with replicas */
+	ZPOOL_STATUS_MISSING_DEV_NR,	/* missing device with no replicas */
+	ZPOOL_STATUS_CORRUPT_LABEL_R,	/* bad device label with replicas */
+	ZPOOL_STATUS_CORRUPT_LABEL_NR,	/* bad device label with no replicas */
+	ZPOOL_STATUS_BAD_GUID_SUM,	/* sum of device guids didn't match */
+	ZPOOL_STATUS_CORRUPT_POOL,	/* pool metadata is corrupted */
+	ZPOOL_STATUS_CORRUPT_DATA,	/* data errors in user (meta)data */
+	ZPOOL_STATUS_FAILING_DEV,	/* device experiencing errors */
+	ZPOOL_STATUS_VERSION_NEWER,	/* newer on-disk version */
+	ZPOOL_STATUS_HOSTID_MISMATCH,	/* last accessed by another system */
+	ZPOOL_STATUS_IO_FAILURE_WAIT,	/* failed I/O, failmode 'wait' */
+	ZPOOL_STATUS_IO_FAILURE_CONTINUE, /* failed I/O, failmode 'continue' */
+	ZPOOL_STATUS_BAD_LOG,		/* cannot read log chain(s) */
+
+	/*
+	 * If the pool has unsupported features but can still be opened in
+	 * read-only mode, its status is ZPOOL_STATUS_UNSUP_FEAT_WRITE. If the
+	 * pool has unsupported features but cannot be opened at all, its
+	 * status is ZPOOL_STATUS_UNSUP_FEAT_READ.
+	 */
+	ZPOOL_STATUS_UNSUP_FEAT_READ,	/* unsupported features for read */
+	ZPOOL_STATUS_UNSUP_FEAT_WRITE,	/* unsupported features for write */
+
+	/*
+	 * These faults have no corresponding message ID.  At the time we are
+	 * checking the status, the original reason for the FMA fault (I/O or
+	 * checksum errors) has been lost.
+	 */
+	ZPOOL_STATUS_FAULTED_DEV_R,	/* faulted device with replicas */
+	ZPOOL_STATUS_FAULTED_DEV_NR,	/* faulted device with no replicas */
+
+	/*
+	 * The following are not faults per se, but still an error possibly
+	 * requiring administrative attention.  There is no corresponding
+	 * message ID.
+	 */
+	ZPOOL_STATUS_VERSION_OLDER,	/* older on-disk version */
+	ZPOOL_STATUS_RESILVERING,	/* device being resilvered */
+	ZPOOL_STATUS_OFFLINE_DEV,	/* device online */
+	ZPOOL_STATUS_REMOVED_DEV,	/* removed device */
+
+	/*
+	 * Finally, the following indicates a healthy pool.
+	 */
+	ZPOOL_STATUS_OK
+} zpool_status_t;
+
+extern zpool_status_t zpool_get_status(zpool_handle_t *, char **);
+extern zpool_status_t zpool_import_status(nvlist_t *, char **);
+extern void zpool_dump_ddt(const ddt_stat_t *dds, const ddt_histogram_t *ddh);
+
+/*
+ * Statistics and configuration functions.
+ */
+extern nvlist_t *zpool_get_config(zpool_handle_t *, nvlist_t **);
+extern nvlist_t *zpool_get_features(zpool_handle_t *);
+extern int zpool_refresh_stats(zpool_handle_t *, boolean_t *);
+extern int zpool_get_errlog(zpool_handle_t *, nvlist_t **);
+
+/*
+ * Import and export functions
+ */
+extern int zpool_export(zpool_handle_t *, boolean_t);
+extern int zpool_export_force(zpool_handle_t *);
+extern int zpool_import(libzfs_handle_t *, nvlist_t *, const char *,
+    char *altroot);
+extern int zpool_import_props(libzfs_handle_t *, nvlist_t *, const char *,
+    nvlist_t *, int);
+extern void zpool_print_unsup_feat(nvlist_t *config);
+
+/*
+ * Search for pools to import
+ */
+
+typedef struct importargs {
+	char **path;		/* a list of paths to search		*/
+	int paths;		/* number of paths to search		*/
+	char *poolname;		/* name of a pool to find		*/
+	uint64_t guid;		/* guid of a pool to find		*/
+	char *cachefile;	/* cachefile to use for import		*/
+	int can_be_active : 1;	/* can the pool be active?		*/
+	int unique : 1;		/* does 'poolname' already exist?	*/
+	int exists : 1;		/* set on return if pool already exists	*/
+} importargs_t;
+
+extern nvlist_t *zpool_search_import(libzfs_handle_t *, importargs_t *);
+
+/* legacy pool search routines */
+extern nvlist_t *zpool_find_import(libzfs_handle_t *, int, char **);
+extern nvlist_t *zpool_find_import_cached(libzfs_handle_t *, const char *,
+    char *, uint64_t);
+
+/*
+ * Miscellaneous pool functions
+ */
+struct zfs_cmd;
+
+extern const char *zfs_history_event_names[LOG_END];
+
+extern char *zpool_vdev_name(libzfs_handle_t *, zpool_handle_t *, nvlist_t *,
+    boolean_t verbose);
+extern int zpool_upgrade(zpool_handle_t *, uint64_t);
+extern int zpool_get_history(zpool_handle_t *, nvlist_t **);
+extern int zpool_history_unpack(char *, uint64_t, uint64_t *,
+    nvlist_t ***, uint_t *);
+extern void zpool_set_history_str(const char *subcommand, int argc,
+    char **argv, char *history_str);
+extern int zpool_stage_history(libzfs_handle_t *, const char *);
+extern void zpool_obj_to_path(zpool_handle_t *, uint64_t, uint64_t, char *,
+    size_t len);
+extern int zfs_ioctl(libzfs_handle_t *, int, struct zfs_cmd *);
+extern int zpool_get_physpath(zpool_handle_t *, char *, size_t);
+extern void zpool_explain_recover(libzfs_handle_t *, const char *, int,
+    nvlist_t *);
+
+/*
+ * Basic handle manipulations.  These functions do not create or destroy the
+ * underlying datasets, only the references to them.
+ */
+extern zfs_handle_t *zfs_open(libzfs_handle_t *, const char *, int);
+extern zfs_handle_t *zfs_handle_dup(zfs_handle_t *);
+extern void zfs_close(zfs_handle_t *);
+extern zfs_type_t zfs_get_type(const zfs_handle_t *);
+extern const char *zfs_get_name(const zfs_handle_t *);
+extern zpool_handle_t *zfs_get_pool_handle(const zfs_handle_t *);
+
+/*
+ * Property management functions.  Some functions are shared with the kernel,
+ * and are found in sys/fs/zfs.h.
+ */
+
+/*
+ * zfs dataset property management
+ */
+extern const char *zfs_prop_default_string(zfs_prop_t);
+extern uint64_t zfs_prop_default_numeric(zfs_prop_t);
+extern const char *zfs_prop_column_name(zfs_prop_t);
+extern boolean_t zfs_prop_align_right(zfs_prop_t);
+
+extern nvlist_t *zfs_valid_proplist(libzfs_handle_t *, zfs_type_t,
+    nvlist_t *, uint64_t, zfs_handle_t *, const char *);
+
+extern const char *zfs_prop_to_name(zfs_prop_t);
+extern int zfs_prop_set(zfs_handle_t *, const char *, const char *);
+extern int zfs_prop_get(zfs_handle_t *, zfs_prop_t, char *, size_t,
+    zprop_source_t *, char *, size_t, boolean_t);
+extern int zfs_prop_get_recvd(zfs_handle_t *, const char *, char *, size_t,
+    boolean_t);
+extern int zfs_prop_get_numeric(zfs_handle_t *, zfs_prop_t, uint64_t *,
+    zprop_source_t *, char *, size_t);
+extern int zfs_prop_get_userquota_int(zfs_handle_t *zhp, const char *propname,
+    uint64_t *propvalue);
+extern int zfs_prop_get_userquota(zfs_handle_t *zhp, const char *propname,
+    char *propbuf, int proplen, boolean_t literal);
+extern int zfs_prop_get_written_int(zfs_handle_t *zhp, const char *propname,
+    uint64_t *propvalue);
+extern int zfs_prop_get_written(zfs_handle_t *zhp, const char *propname,
+    char *propbuf, int proplen, boolean_t literal);
+extern int zfs_prop_get_feature(zfs_handle_t *zhp, const char *propname,
+    char *buf, size_t len);
+extern int zfs_get_snapused_int(zfs_handle_t *firstsnap, zfs_handle_t *lastsnap,
+    uint64_t *usedp);
+extern uint64_t zfs_prop_get_int(zfs_handle_t *, zfs_prop_t);
+extern int zfs_prop_inherit(zfs_handle_t *, const char *, boolean_t);
+extern const char *zfs_prop_values(zfs_prop_t);
+extern int zfs_prop_is_string(zfs_prop_t prop);
+extern nvlist_t *zfs_get_user_props(zfs_handle_t *);
+extern nvlist_t *zfs_get_recvd_props(zfs_handle_t *);
+extern nvlist_t *zfs_get_clones_nvl(zfs_handle_t *);
+
+
+typedef struct zprop_list {
+	int		pl_prop;
+	char		*pl_user_prop;
+	struct zprop_list *pl_next;
+	boolean_t	pl_all;
+	size_t		pl_width;
+	size_t		pl_recvd_width;
+	boolean_t	pl_fixed;
+} zprop_list_t;
+
+extern int zfs_expand_proplist(zfs_handle_t *, zprop_list_t **, boolean_t);
+extern void zfs_prune_proplist(zfs_handle_t *, uint8_t *);
+
+#define	ZFS_MOUNTPOINT_NONE	"none"
+#define	ZFS_MOUNTPOINT_LEGACY	"legacy"
+
+#define	ZFS_FEATURE_DISABLED	"disabled"
+#define	ZFS_FEATURE_ENABLED	"enabled"
+#define	ZFS_FEATURE_ACTIVE	"active"
+
+#define	ZFS_UNSUPPORTED_INACTIVE	"inactive"
+#define	ZFS_UNSUPPORTED_READONLY	"readonly"
+
+/*
+ * zpool property management
+ */
+extern int zpool_expand_proplist(zpool_handle_t *, zprop_list_t **);
+extern int zpool_prop_get_feature(zpool_handle_t *, const char *, char *,
+    size_t);
+extern const char *zpool_prop_default_string(zpool_prop_t);
+extern uint64_t zpool_prop_default_numeric(zpool_prop_t);
+extern const char *zpool_prop_column_name(zpool_prop_t);
+extern boolean_t zpool_prop_align_right(zpool_prop_t);
+
+/*
+ * Functions shared by zfs and zpool property management.
+ */
+extern int zprop_iter(zprop_func func, void *cb, boolean_t show_all,
+    boolean_t ordered, zfs_type_t type);
+extern int zprop_get_list(libzfs_handle_t *, char *, zprop_list_t **,
+    zfs_type_t);
+extern void zprop_free_list(zprop_list_t *);
+
+#define	ZFS_GET_NCOLS	5
+
+typedef enum {
+	GET_COL_NONE,
+	GET_COL_NAME,
+	GET_COL_PROPERTY,
+	GET_COL_VALUE,
+	GET_COL_RECVD,
+	GET_COL_SOURCE
+} zfs_get_column_t;
+
+/*
+ * Functions for printing zfs or zpool properties
+ */
+typedef struct zprop_get_cbdata {
+	int cb_sources;
+	zfs_get_column_t cb_columns[ZFS_GET_NCOLS];
+	int cb_colwidths[ZFS_GET_NCOLS + 1];
+	boolean_t cb_scripted;
+	boolean_t cb_literal;
+	boolean_t cb_first;
+	zprop_list_t *cb_proplist;
+	zfs_type_t cb_type;
+} zprop_get_cbdata_t;
+
+void zprop_print_one_property(const char *, zprop_get_cbdata_t *,
+    const char *, const char *, zprop_source_t, const char *,
+    const char *);
+
+/*
+ * Iterator functions.
+ */
+typedef int (*zfs_iter_f)(zfs_handle_t *, void *);
+extern int zfs_iter_root(libzfs_handle_t *, zfs_iter_f, void *);
+extern int zfs_iter_children(zfs_handle_t *, zfs_iter_f, void *);
+extern int zfs_iter_dependents(zfs_handle_t *, boolean_t, zfs_iter_f, void *);
+extern int zfs_iter_filesystems(zfs_handle_t *, zfs_iter_f, void *);
+extern int zfs_iter_snapshots(zfs_handle_t *, zfs_iter_f, void *);
+extern int zfs_iter_snapshots_sorted(zfs_handle_t *, zfs_iter_f, void *);
+extern int zfs_iter_snapspec(zfs_handle_t *, const char *, zfs_iter_f, void *);
+
+typedef struct get_all_cb {
+	zfs_handle_t	**cb_handles;
+	size_t		cb_alloc;
+	size_t		cb_used;
+	boolean_t	cb_verbose;
+	int		(*cb_getone)(zfs_handle_t *, void *);
+} get_all_cb_t;
+
+void libzfs_add_handle(get_all_cb_t *, zfs_handle_t *);
+int libzfs_dataset_cmp(const void *, const void *);
+
+/*
+ * Functions to create and destroy datasets.
+ */
+extern int zfs_create(libzfs_handle_t *, const char *, zfs_type_t,
+    nvlist_t *);
+extern int zfs_create_ancestors(libzfs_handle_t *, const char *);
+extern int zfs_destroy(zfs_handle_t *, boolean_t);
+extern int zfs_destroy_snaps(zfs_handle_t *, char *, boolean_t);
+extern int zfs_destroy_snaps_nvl(zfs_handle_t *, nvlist_t *, boolean_t);
+extern int zfs_clone(zfs_handle_t *, const char *, nvlist_t *);
+extern int zfs_snapshot(libzfs_handle_t *, const char *, boolean_t, nvlist_t *);
+extern int zfs_rollback(zfs_handle_t *, zfs_handle_t *, boolean_t);
+extern int zfs_rename(zfs_handle_t *, const char *, boolean_t, boolean_t);
+
+typedef struct sendflags {
+	/* print informational messages (ie, -v was specified) */
+	boolean_t verbose;
+
+	/* recursive send  (ie, -R) */
+	boolean_t replicate;
+
+	/* for incrementals, do all intermediate snapshots */
+	boolean_t doall;
+
+	/* if dataset is a clone, do incremental from its origin */
+	boolean_t fromorigin;
+
+	/* do deduplication */
+	boolean_t dedup;
+
+	/* send properties (ie, -p) */
+	boolean_t props;
+
+	/* do not send (no-op, ie. -n) */
+	boolean_t dryrun;
+
+	/* parsable verbose output (ie. -P) */
+	boolean_t parsable;
+
+	/* show progress (ie. -v) */
+	boolean_t progress;
+} sendflags_t;
+
+typedef boolean_t (snapfilter_cb_t)(zfs_handle_t *, void *);
+
+extern int zfs_send(zfs_handle_t *, const char *, const char *,
+    sendflags_t *, int, snapfilter_cb_t, void *, nvlist_t **);
+
+extern int zfs_promote(zfs_handle_t *);
+extern int zfs_hold(zfs_handle_t *, const char *, const char *, boolean_t,
+    boolean_t, boolean_t, int, uint64_t, uint64_t);
+extern int zfs_release(zfs_handle_t *, const char *, const char *, boolean_t);
+extern int zfs_get_holds(zfs_handle_t *, nvlist_t **);
+extern uint64_t zvol_volsize_to_reservation(uint64_t, nvlist_t *);
+
+typedef int (*zfs_userspace_cb_t)(void *arg, const char *domain,
+    uid_t rid, uint64_t space);
+
+extern int zfs_userspace(zfs_handle_t *, zfs_userquota_prop_t,
+    zfs_userspace_cb_t, void *);
+
+extern int zfs_get_fsacl(zfs_handle_t *, nvlist_t **);
+extern int zfs_set_fsacl(zfs_handle_t *, boolean_t, nvlist_t *);
+
+typedef struct recvflags {
+	/* print informational messages (ie, -v was specified) */
+	boolean_t verbose;
+
+	/* the destination is a prefix, not the exact fs (ie, -d) */
+	boolean_t isprefix;
+
+	/*
+	 * Only the tail of the sent snapshot path is appended to the
+	 * destination to determine the received snapshot name (ie, -e).
+	 */
+	boolean_t istail;
+
+	/* do not actually do the recv, just check if it would work (ie, -n) */
+	boolean_t dryrun;
+
+	/* rollback/destroy filesystems as necessary (eg, -F) */
+	boolean_t force;
+
+	/* set "canmount=off" on all modified filesystems */
+	boolean_t canmountoff;
+
+	/* byteswap flag is used internally; callers need not specify */
+	boolean_t byteswap;
+
+	/* do not mount file systems as they are extracted (private) */
+	boolean_t nomount;
+} recvflags_t;
+
+extern int zfs_receive(libzfs_handle_t *, const char *, recvflags_t *,
+    int, avl_tree_t *);
+
+typedef enum diff_flags {
+	ZFS_DIFF_PARSEABLE = 0x1,
+	ZFS_DIFF_TIMESTAMP = 0x2,
+	ZFS_DIFF_CLASSIFY = 0x4
+} diff_flags_t;
+
+extern int zfs_show_diffs(zfs_handle_t *, int, const char *, const char *,
+    int);
+
+/*
+ * Miscellaneous functions.
+ */
+extern const char *zfs_type_to_name(zfs_type_t);
+extern void zfs_refresh_properties(zfs_handle_t *);
+extern int zfs_name_valid(const char *, zfs_type_t);
+extern zfs_handle_t *zfs_path_to_zhandle(libzfs_handle_t *, char *, zfs_type_t);
+extern boolean_t zfs_dataset_exists(libzfs_handle_t *, const char *,
+    zfs_type_t);
+extern int zfs_spa_version(zfs_handle_t *, int *);
+
+/*
+ * Mount support functions.
+ */
+extern boolean_t is_mounted(libzfs_handle_t *, const char *special, char **);
+extern boolean_t zfs_is_mounted(zfs_handle_t *, char **);
+extern int zfs_mount(zfs_handle_t *, const char *, int);
+extern int zfs_unmount(zfs_handle_t *, const char *, int);
+extern int zfs_unmountall(zfs_handle_t *, int);
+
+/*
+ * Share support functions.
+ */
+extern boolean_t zfs_is_shared(zfs_handle_t *);
+extern int zfs_share(zfs_handle_t *);
+extern int zfs_unshare(zfs_handle_t *);
+
+/*
+ * Protocol-specific share support functions.
+ */
+extern boolean_t zfs_is_shared_nfs(zfs_handle_t *, char **);
+extern boolean_t zfs_is_shared_smb(zfs_handle_t *, char **);
+extern int zfs_share_nfs(zfs_handle_t *);
+extern int zfs_share_smb(zfs_handle_t *);
+extern int zfs_shareall(zfs_handle_t *);
+extern int zfs_unshare_nfs(zfs_handle_t *, const char *);
+extern int zfs_unshare_smb(zfs_handle_t *, const char *);
+extern int zfs_unshareall_nfs(zfs_handle_t *);
+extern int zfs_unshareall_smb(zfs_handle_t *);
+extern int zfs_unshareall_bypath(zfs_handle_t *, const char *);
+extern int zfs_unshareall(zfs_handle_t *);
+extern int zfs_deleg_share_nfs(libzfs_handle_t *, char *, char *, char *,
+    void *, void *, int, zfs_share_op_t);
+
+/*
+ * When dealing with nvlists, verify() is extremely useful
+ */
+#ifdef NDEBUG
+#define	verify(EX)	((void)(EX))
+#else
+#define	verify(EX)	assert(EX)
+#endif
+
+/*
+ * Utility function to convert a number to a human-readable form.
+ */
+extern void zfs_nicenum(uint64_t, char *, size_t);
+extern int zfs_nicestrtonum(libzfs_handle_t *, const char *, uint64_t *);
+
+/*
+ * Given a device or file, determine if it is part of a pool.
+ */
+extern int zpool_in_use(libzfs_handle_t *, int, pool_state_t *, char **,
+    boolean_t *);
+
+/*
+ * Label manipulation.
+ */
+extern int zpool_read_label(int, nvlist_t **);
+extern int zpool_clear_label(int);
+
+/* is this zvol valid for use as a dump device? */
+extern int zvol_check_dump_config(char *);
+
+/*
+ * Management interfaces for SMB ACL files
+ */
+
+int zfs_smb_acl_add(libzfs_handle_t *, char *, char *, char *);
+int zfs_smb_acl_remove(libzfs_handle_t *, char *, char *, char *);
+int zfs_smb_acl_purge(libzfs_handle_t *, char *, char *);
+int zfs_smb_acl_rename(libzfs_handle_t *, char *, char *, char *, char *);
+
+/*
+ * Enable and disable datasets within a pool by mounting/unmounting and
+ * sharing/unsharing them.
+ */
+extern int zpool_enable_datasets(zpool_handle_t *, const char *, int);
+extern int zpool_disable_datasets(zpool_handle_t *, boolean_t);
+
+/*
+ * Mappings between vdev and FRU.
+ */
+extern void libzfs_fru_refresh(libzfs_handle_t *);
+extern const char *libzfs_fru_lookup(libzfs_handle_t *, const char *);
+extern const char *libzfs_fru_devpath(libzfs_handle_t *, const char *);
+extern boolean_t libzfs_fru_compare(libzfs_handle_t *, const char *,
+    const char *);
+extern boolean_t libzfs_fru_notself(libzfs_handle_t *, const char *);
+extern int zpool_fru_set(zpool_handle_t *, uint64_t, const char *);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _LIBZFS_H */

lib/libzfs/common/libzfs_changelist.c

@@ -0,0 +1,693 @@
+/*
+ * 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 2010 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ *
+ * Portions Copyright 2007 Ramprakash Jelari
+ */
+
+#include <libintl.h>
+#include <libuutil.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <zone.h>
+
+#include <libzfs.h>
+
+#include "libzfs_impl.h"
+
+/*
+ * Structure to keep track of dataset state.  Before changing the 'sharenfs' or
+ * 'mountpoint' property, we record whether the filesystem was previously
+ * mounted/shared.  This prior state dictates whether we remount/reshare the
+ * dataset after the property has been changed.
+ *
+ * The interface consists of the following sequence of functions:
+ *
+ * 	changelist_gather()
+ * 	changelist_prefix()
+ * 	< change property >
+ * 	changelist_postfix()
+ * 	changelist_free()
+ *
+ * Other interfaces:
+ *
+ * changelist_remove() - remove a node from a gathered list
+ * changelist_rename() - renames all datasets appropriately when doing a rename
+ * changelist_unshare() - unshares all the nodes in a given changelist
+ * changelist_haszonedchild() - check if there is any child exported to
+ *				a local zone
+ */
+typedef struct prop_changenode {
+	zfs_handle_t		*cn_handle;
+	int			cn_shared;
+	int			cn_mounted;
+	int			cn_zoned;
+	boolean_t		cn_needpost;	/* is postfix() needed? */
+	uu_list_node_t		cn_listnode;
+} prop_changenode_t;
+
+struct prop_changelist {
+	zfs_prop_t		cl_prop;
+	zfs_prop_t		cl_realprop;
+	zfs_prop_t		cl_shareprop;  /* used with sharenfs/sharesmb */
+	uu_list_pool_t		*cl_pool;
+	uu_list_t		*cl_list;
+	boolean_t		cl_waslegacy;
+	boolean_t		cl_allchildren;
+	boolean_t		cl_alldependents;
+	int			cl_mflags;	/* Mount flags */
+	int			cl_gflags;	/* Gather request flags */
+	boolean_t		cl_haszonedchild;
+	boolean_t		cl_sorted;
+};
+
+/*
+ * If the property is 'mountpoint', go through and unmount filesystems as
+ * necessary.  We don't do the same for 'sharenfs', because we can just re-share
+ * with different options without interrupting service. We do handle 'sharesmb'
+ * since there may be old resource names that need to be removed.
+ */
+int
+changelist_prefix(prop_changelist_t *clp)
+{
+	prop_changenode_t *cn;
+	int ret = 0;
+
+	if (clp->cl_prop != ZFS_PROP_MOUNTPOINT &&
+	    clp->cl_prop != ZFS_PROP_SHARESMB)
+		return (0);
+
+	for (cn = uu_list_first(clp->cl_list); cn != NULL;
+	    cn = uu_list_next(clp->cl_list, cn)) {
+
+		/* if a previous loop failed, set the remaining to false */
+		if (ret == -1) {
+			cn->cn_needpost = B_FALSE;
+			continue;
+		}
+
+		/*
+		 * If we are in the global zone, but this dataset is exported
+		 * to a local zone, do nothing.
+		 */
+		if (getzoneid() == GLOBAL_ZONEID && cn->cn_zoned)
+			continue;
+
+		if (!ZFS_IS_VOLUME(cn->cn_handle)) {
+			/*
+			 * Do the property specific processing.
+			 */
+			switch (clp->cl_prop) {
+			case ZFS_PROP_MOUNTPOINT:
+				if (zfs_unmount(cn->cn_handle, NULL,
+				    clp->cl_mflags) != 0) {
+					ret = -1;
+					cn->cn_needpost = B_FALSE;
+				}
+				break;
+			case ZFS_PROP_SHARESMB:
+				(void) zfs_unshare_smb(cn->cn_handle, NULL);
+				break;
+			}
+		}
+	}
+
+	if (ret == -1)
+		(void) changelist_postfix(clp);
+
+	return (ret);
+}
+
+/*
+ * If the property is 'mountpoint' or 'sharenfs', go through and remount and/or
+ * reshare the filesystems as necessary.  In changelist_gather() we recorded
+ * whether the filesystem was previously shared or mounted.  The action we take
+ * depends on the previous state, and whether the value was previously 'legacy'.
+ * For non-legacy properties, we only remount/reshare the filesystem if it was
+ * previously mounted/shared.  Otherwise, we always remount/reshare the
+ * filesystem.
+ */
+int
+changelist_postfix(prop_changelist_t *clp)
+{
+	prop_changenode_t *cn;
+	char shareopts[ZFS_MAXPROPLEN];
+	int errors = 0;
+	libzfs_handle_t *hdl;
+
+	/*
+	 * If we're changing the mountpoint, attempt to destroy the underlying
+	 * mountpoint.  All other datasets will have inherited from this dataset
+	 * (in which case their mountpoints exist in the filesystem in the new
+	 * location), or have explicit mountpoints set (in which case they won't
+	 * be in the changelist).
+	 */
+	if ((cn = uu_list_last(clp->cl_list)) == NULL)
+		return (0);
+
+	if (clp->cl_prop == ZFS_PROP_MOUNTPOINT)
+		remove_mountpoint(cn->cn_handle);
+
+	/*
+	 * It is possible that the changelist_prefix() used libshare
+	 * to unshare some entries. Since libshare caches data, an
+	 * attempt to reshare during postfix can fail unless libshare
+	 * is uninitialized here so that it will reinitialize later.
+	 */
+	if (cn->cn_handle != NULL) {
+		hdl = cn->cn_handle->zfs_hdl;
+		assert(hdl != NULL);
+		zfs_uninit_libshare(hdl);
+	}
+
+	/*
+	 * We walk the datasets in reverse, because we want to mount any parent
+	 * datasets before mounting the children.  We walk all datasets even if
+	 * there are errors.
+	 */
+	for (cn = uu_list_last(clp->cl_list); cn != NULL;
+	    cn = uu_list_prev(clp->cl_list, cn)) {
+
+		boolean_t sharenfs;
+		boolean_t sharesmb;
+		boolean_t mounted;
+
+		/*
+		 * If we are in the global zone, but this dataset is exported
+		 * to a local zone, do nothing.
+		 */
+		if (getzoneid() == GLOBAL_ZONEID && cn->cn_zoned)
+			continue;
+
+		/* Only do post-processing if it's required */
+		if (!cn->cn_needpost)
+			continue;
+		cn->cn_needpost = B_FALSE;
+
+		zfs_refresh_properties(cn->cn_handle);
+
+		if (ZFS_IS_VOLUME(cn->cn_handle))
+			continue;
+
+		/*
+		 * Remount if previously mounted or mountpoint was legacy,
+		 * or sharenfs or sharesmb  property is set.
+		 */
+		sharenfs = ((zfs_prop_get(cn->cn_handle, ZFS_PROP_SHARENFS,
+		    shareopts, sizeof (shareopts), NULL, NULL, 0,
+		    B_FALSE) == 0) && (strcmp(shareopts, "off") != 0));
+
+		sharesmb = ((zfs_prop_get(cn->cn_handle, ZFS_PROP_SHARESMB,
+		    shareopts, sizeof (shareopts), NULL, NULL, 0,
+		    B_FALSE) == 0) && (strcmp(shareopts, "off") != 0));
+
+		mounted = zfs_is_mounted(cn->cn_handle, NULL);
+
+		if (!mounted && (cn->cn_mounted ||
+		    ((sharenfs || sharesmb || clp->cl_waslegacy) &&
+		    (zfs_prop_get_int(cn->cn_handle,
+		    ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_ON)))) {
+
+			if (zfs_mount(cn->cn_handle, NULL, 0) != 0)
+				errors++;
+			else
+				mounted = TRUE;
+		}
+
+		/*
+		 * If the file system is mounted we always re-share even
+		 * if the filesystem is currently shared, so that we can
+		 * adopt any new options.
+		 */
+		if (sharenfs && mounted)
+			errors += zfs_share_nfs(cn->cn_handle);
+		else if (cn->cn_shared || clp->cl_waslegacy)
+			errors += zfs_unshare_nfs(cn->cn_handle, NULL);
+		if (sharesmb && mounted)
+			errors += zfs_share_smb(cn->cn_handle);
+		else if (cn->cn_shared || clp->cl_waslegacy)
+			errors += zfs_unshare_smb(cn->cn_handle, NULL);
+	}
+
+	return (errors ? -1 : 0);
+}
+
+/*
+ * Is this "dataset" a child of "parent"?
+ */
+boolean_t
+isa_child_of(const char *dataset, const char *parent)
+{
+	int len;
+
+	len = strlen(parent);
+
+	if (strncmp(dataset, parent, len) == 0 &&
+	    (dataset[len] == '@' || dataset[len] == '/' ||
+	    dataset[len] == '\0'))
+		return (B_TRUE);
+	else
+		return (B_FALSE);
+
+}
+
+/*
+ * If we rename a filesystem, child filesystem handles are no longer valid
+ * since we identify each dataset by its name in the ZFS namespace.  As a
+ * result, we have to go through and fix up all the names appropriately.  We
+ * could do this automatically if libzfs kept track of all open handles, but
+ * this is a lot less work.
+ */
+void
+changelist_rename(prop_changelist_t *clp, const char *src, const char *dst)
+{
+	prop_changenode_t *cn;
+	char newname[ZFS_MAXNAMELEN];
+
+	for (cn = uu_list_first(clp->cl_list); cn != NULL;
+	    cn = uu_list_next(clp->cl_list, cn)) {
+		/*
+		 * Do not rename a clone that's not in the source hierarchy.
+		 */
+		if (!isa_child_of(cn->cn_handle->zfs_name, src))
+			continue;
+
+		/*
+		 * Destroy the previous mountpoint if needed.
+		 */
+		remove_mountpoint(cn->cn_handle);
+
+		(void) strlcpy(newname, dst, sizeof (newname));
+		(void) strcat(newname, cn->cn_handle->zfs_name + strlen(src));
+
+		(void) strlcpy(cn->cn_handle->zfs_name, newname,
+		    sizeof (cn->cn_handle->zfs_name));
+	}
+}
+
+/*
+ * Given a gathered changelist for the 'sharenfs' or 'sharesmb' property,
+ * unshare all the datasets in the list.
+ */
+int
+changelist_unshare(prop_changelist_t *clp, zfs_share_proto_t *proto)
+{
+	prop_changenode_t *cn;
+	int ret = 0;
+
+	if (clp->cl_prop != ZFS_PROP_SHARENFS &&
+	    clp->cl_prop != ZFS_PROP_SHARESMB)
+		return (0);
+
+	for (cn = uu_list_first(clp->cl_list); cn != NULL;
+	    cn = uu_list_next(clp->cl_list, cn)) {
+		if (zfs_unshare_proto(cn->cn_handle, NULL, proto) != 0)
+			ret = -1;
+	}
+
+	return (ret);
+}
+
+/*
+ * Check if there is any child exported to a local zone in a given changelist.
+ * This information has already been recorded while gathering the changelist
+ * via changelist_gather().
+ */
+int
+changelist_haszonedchild(prop_changelist_t *clp)
+{
+	return (clp->cl_haszonedchild);
+}
+
+/*
+ * Remove a node from a gathered list.
+ */
+void
+changelist_remove(prop_changelist_t *clp, const char *name)
+{
+	prop_changenode_t *cn;
+
+	for (cn = uu_list_first(clp->cl_list); cn != NULL;
+	    cn = uu_list_next(clp->cl_list, cn)) {
+
+		if (strcmp(cn->cn_handle->zfs_name, name) == 0) {
+			uu_list_remove(clp->cl_list, cn);
+			zfs_close(cn->cn_handle);
+			free(cn);
+			return;
+		}
+	}
+}
+
+/*
+ * Release any memory associated with a changelist.
+ */
+void
+changelist_free(prop_changelist_t *clp)
+{
+	prop_changenode_t *cn;
+	void *cookie;
+
+	if (clp->cl_list) {
+		cookie = NULL;
+		while ((cn = uu_list_teardown(clp->cl_list, &cookie)) != NULL) {
+			zfs_close(cn->cn_handle);
+			free(cn);
+		}
+
+		uu_list_destroy(clp->cl_list);
+	}
+	if (clp->cl_pool)
+		uu_list_pool_destroy(clp->cl_pool);
+
+	free(clp);
+}
+
+static int
+change_one(zfs_handle_t *zhp, void *data)
+{
+	prop_changelist_t *clp = data;
+	char property[ZFS_MAXPROPLEN];
+	char where[64];
+	prop_changenode_t *cn;
+	zprop_source_t sourcetype;
+	zprop_source_t share_sourcetype;
+
+	/*
+	 * We only want to unmount/unshare those filesystems that may inherit
+	 * from the target filesystem.  If we find any filesystem with a
+	 * locally set mountpoint, we ignore any children since changing the
+	 * property will not affect them.  If this is a rename, we iterate
+	 * over all children regardless, since we need them unmounted in
+	 * order to do the rename.  Also, if this is a volume and we're doing
+	 * a rename, then always add it to the changelist.
+	 */
+
+	if (!(ZFS_IS_VOLUME(zhp) && clp->cl_realprop == ZFS_PROP_NAME) &&
+	    zfs_prop_get(zhp, clp->cl_prop, property,
+	    sizeof (property), &sourcetype, where, sizeof (where),
+	    B_FALSE) != 0) {
+		zfs_close(zhp);
+		return (0);
+	}
+
+	/*
+	 * If we are "watching" sharenfs or sharesmb
+	 * then check out the companion property which is tracked
+	 * in cl_shareprop
+	 */
+	if (clp->cl_shareprop != ZPROP_INVAL &&
+	    zfs_prop_get(zhp, clp->cl_shareprop, property,
+	    sizeof (property), &share_sourcetype, where, sizeof (where),
+	    B_FALSE) != 0) {
+		zfs_close(zhp);
+		return (0);
+	}
+
+	if (clp->cl_alldependents || clp->cl_allchildren ||
+	    sourcetype == ZPROP_SRC_DEFAULT ||
+	    sourcetype == ZPROP_SRC_INHERITED ||
+	    (clp->cl_shareprop != ZPROP_INVAL &&
+	    (share_sourcetype == ZPROP_SRC_DEFAULT ||
+	    share_sourcetype == ZPROP_SRC_INHERITED))) {
+		if ((cn = zfs_alloc(zfs_get_handle(zhp),
+		    sizeof (prop_changenode_t))) == NULL) {
+			zfs_close(zhp);
+			return (-1);
+		}
+
+		cn->cn_handle = zhp;
+		cn->cn_mounted = (clp->cl_gflags & CL_GATHER_MOUNT_ALWAYS) ||
+		    zfs_is_mounted(zhp, NULL);
+		cn->cn_shared = zfs_is_shared(zhp);
+		cn->cn_zoned = zfs_prop_get_int(zhp, ZFS_PROP_ZONED);
+		cn->cn_needpost = B_TRUE;
+
+		/* Indicate if any child is exported to a local zone. */
+		if (getzoneid() == GLOBAL_ZONEID && cn->cn_zoned)
+			clp->cl_haszonedchild = B_TRUE;
+
+		uu_list_node_init(cn, &cn->cn_listnode, clp->cl_pool);
+
+		if (clp->cl_sorted) {
+			uu_list_index_t idx;
+
+			(void) uu_list_find(clp->cl_list, cn, NULL,
+			    &idx);
+			uu_list_insert(clp->cl_list, cn, idx);
+		} else {
+			/*
+			 * Add this child to beginning of the list. Children
+			 * below this one in the hierarchy will get added above
+			 * this one in the list. This produces a list in
+			 * reverse dataset name order.
+			 * This is necessary when the original mountpoint
+			 * is legacy or none.
+			 */
+			ASSERT(!clp->cl_alldependents);
+			verify(uu_list_insert_before(clp->cl_list,
+			    uu_list_first(clp->cl_list), cn) == 0);
+		}
+
+		if (!clp->cl_alldependents)
+			return (zfs_iter_children(zhp, change_one, data));
+	} else {
+		zfs_close(zhp);
+	}
+
+	return (0);
+}
+
+/*ARGSUSED*/
+static int
+compare_mountpoints(const void *a, const void *b, void *unused)
+{
+	const prop_changenode_t *ca = a;
+	const prop_changenode_t *cb = b;
+
+	char mounta[MAXPATHLEN];
+	char mountb[MAXPATHLEN];
+
+	boolean_t hasmounta, hasmountb;
+
+	/*
+	 * When unsharing or unmounting filesystems, we need to do it in
+	 * mountpoint order.  This allows the user to have a mountpoint
+	 * hierarchy that is different from the dataset hierarchy, and still
+	 * allow it to be changed.  However, if either dataset doesn't have a
+	 * mountpoint (because it is a volume or a snapshot), we place it at the
+	 * end of the list, because it doesn't affect our change at all.
+	 */
+	hasmounta = (zfs_prop_get(ca->cn_handle, ZFS_PROP_MOUNTPOINT, mounta,
+	    sizeof (mounta), NULL, NULL, 0, B_FALSE) == 0);
+	hasmountb = (zfs_prop_get(cb->cn_handle, ZFS_PROP_MOUNTPOINT, mountb,
+	    sizeof (mountb), NULL, NULL, 0, B_FALSE) == 0);
+
+	if (!hasmounta && hasmountb)
+		return (-1);
+	else if (hasmounta && !hasmountb)
+		return (1);
+	else if (!hasmounta && !hasmountb)
+		return (0);
+	else
+		return (strcmp(mountb, mounta));
+}
+
+/*
+ * Given a ZFS handle and a property, construct a complete list of datasets
+ * that need to be modified as part of this process.  For anything but the
+ * 'mountpoint' and 'sharenfs' properties, this just returns an empty list.
+ * Otherwise, we iterate over all children and look for any datasets that
+ * inherit the property.  For each such dataset, we add it to the list and
+ * mark whether it was shared beforehand.
+ */
+prop_changelist_t *
+changelist_gather(zfs_handle_t *zhp, zfs_prop_t prop, int gather_flags,
+    int mnt_flags)
+{
+	prop_changelist_t *clp;
+	prop_changenode_t *cn;
+	zfs_handle_t *temp;
+	char property[ZFS_MAXPROPLEN];
+	uu_compare_fn_t *compare = NULL;
+	boolean_t legacy = B_FALSE;
+
+	if ((clp = zfs_alloc(zhp->zfs_hdl, sizeof (prop_changelist_t))) == NULL)
+		return (NULL);
+
+	/*
+	 * For mountpoint-related tasks, we want to sort everything by
+	 * mountpoint, so that we mount and unmount them in the appropriate
+	 * order, regardless of their position in the hierarchy.
+	 */
+	if (prop == ZFS_PROP_NAME || prop == ZFS_PROP_ZONED ||
+	    prop == ZFS_PROP_MOUNTPOINT || prop == ZFS_PROP_SHARENFS ||
+	    prop == ZFS_PROP_SHARESMB) {
+
+		if (zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT,
+		    property, sizeof (property),
+		    NULL, NULL, 0, B_FALSE) == 0 &&
+		    (strcmp(property, "legacy") == 0 ||
+		    strcmp(property, "none") == 0)) {
+
+			legacy = B_TRUE;
+		}
+		if (!legacy) {
+			compare = compare_mountpoints;
+			clp->cl_sorted = B_TRUE;
+		}
+	}
+
+	clp->cl_pool = uu_list_pool_create("changelist_pool",
+	    sizeof (prop_changenode_t),
+	    offsetof(prop_changenode_t, cn_listnode),
+	    compare, 0);
+	if (clp->cl_pool == NULL) {
+		assert(uu_error() == UU_ERROR_NO_MEMORY);
+		(void) zfs_error(zhp->zfs_hdl, EZFS_NOMEM, "internal error");
+		changelist_free(clp);
+		return (NULL);
+	}
+
+	clp->cl_list = uu_list_create(clp->cl_pool, NULL,
+	    clp->cl_sorted ? UU_LIST_SORTED : 0);
+	clp->cl_gflags = gather_flags;
+	clp->cl_mflags = mnt_flags;
+
+	if (clp->cl_list == NULL) {
+		assert(uu_error() == UU_ERROR_NO_MEMORY);
+		(void) zfs_error(zhp->zfs_hdl, EZFS_NOMEM, "internal error");
+		changelist_free(clp);
+		return (NULL);
+	}
+
+	/*
+	 * If this is a rename or the 'zoned' property, we pretend we're
+	 * changing the mountpoint and flag it so we can catch all children in
+	 * change_one().
+	 *
+	 * Flag cl_alldependents to catch all children plus the dependents
+	 * (clones) that are not in the hierarchy.
+	 */
+	if (prop == ZFS_PROP_NAME) {
+		clp->cl_prop = ZFS_PROP_MOUNTPOINT;
+		clp->cl_alldependents = B_TRUE;
+	} else if (prop == ZFS_PROP_ZONED) {
+		clp->cl_prop = ZFS_PROP_MOUNTPOINT;
+		clp->cl_allchildren = B_TRUE;
+	} else if (prop == ZFS_PROP_CANMOUNT) {
+		clp->cl_prop = ZFS_PROP_MOUNTPOINT;
+	} else if (prop == ZFS_PROP_VOLSIZE) {
+		clp->cl_prop = ZFS_PROP_MOUNTPOINT;
+	} else {
+		clp->cl_prop = prop;
+	}
+	clp->cl_realprop = prop;
+
+	if (clp->cl_prop != ZFS_PROP_MOUNTPOINT &&
+	    clp->cl_prop != ZFS_PROP_SHARENFS &&
+	    clp->cl_prop != ZFS_PROP_SHARESMB)
+		return (clp);
+
+	/*
+	 * If watching SHARENFS or SHARESMB then
+	 * also watch its companion property.
+	 */
+	if (clp->cl_prop == ZFS_PROP_SHARENFS)
+		clp->cl_shareprop = ZFS_PROP_SHARESMB;
+	else if (clp->cl_prop == ZFS_PROP_SHARESMB)
+		clp->cl_shareprop = ZFS_PROP_SHARENFS;
+
+	if (clp->cl_alldependents) {
+		if (zfs_iter_dependents(zhp, B_TRUE, change_one, clp) != 0) {
+			changelist_free(clp);
+			return (NULL);
+		}
+	} else if (zfs_iter_children(zhp, change_one, clp) != 0) {
+		changelist_free(clp);
+		return (NULL);
+	}
+
+	/*
+	 * We have to re-open ourselves because we auto-close all the handles
+	 * and can't tell the difference.
+	 */
+	if ((temp = zfs_open(zhp->zfs_hdl, zfs_get_name(zhp),
+	    ZFS_TYPE_DATASET)) == NULL) {
+		changelist_free(clp);
+		return (NULL);
+	}
+
+	/*
+	 * Always add ourself to the list.  We add ourselves to the end so that
+	 * we're the last to be unmounted.
+	 */
+	if ((cn = zfs_alloc(zhp->zfs_hdl,
+	    sizeof (prop_changenode_t))) == NULL) {
+		zfs_close(temp);
+		changelist_free(clp);
+		return (NULL);
+	}
+
+	cn->cn_handle = temp;
+	cn->cn_mounted = (clp->cl_gflags & CL_GATHER_MOUNT_ALWAYS) ||
+	    zfs_is_mounted(temp, NULL);
+	cn->cn_shared = zfs_is_shared(temp);
+	cn->cn_zoned = zfs_prop_get_int(zhp, ZFS_PROP_ZONED);
+	cn->cn_needpost = B_TRUE;
+
+	uu_list_node_init(cn, &cn->cn_listnode, clp->cl_pool);
+	if (clp->cl_sorted) {
+		uu_list_index_t idx;
+		(void) uu_list_find(clp->cl_list, cn, NULL, &idx);
+		uu_list_insert(clp->cl_list, cn, idx);
+	} else {
+		/*
+		 * Add the target dataset to the end of the list.
+		 * The list is not really unsorted. The list will be
+		 * in reverse dataset name order. This is necessary
+		 * when the original mountpoint is legacy or none.
+		 */
+		verify(uu_list_insert_after(clp->cl_list,
+		    uu_list_last(clp->cl_list), cn) == 0);
+	}
+
+	/*
+	 * If the mountpoint property was previously 'legacy', or 'none',
+	 * record it as the behavior of changelist_postfix() will be different.
+	 */
+	if ((clp->cl_prop == ZFS_PROP_MOUNTPOINT) && legacy) {
+		/*
+		 * do not automatically mount ex-legacy datasets if
+		 * we specifically set canmount to noauto
+		 */
+		if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) !=
+		    ZFS_CANMOUNT_NOAUTO)
+			clp->cl_waslegacy = B_TRUE;
+	}
+
+	return (clp);
+}

lib/libzfs/common/libzfs_config.c

@@ -0,0 +1,405 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+/*
+ * Copyright (c) 2012 by Delphix. All rights reserved.
+ */
+
+/*
+ * The pool configuration repository is stored in /etc/zfs/zpool.cache as a
+ * single packed nvlist.  While it would be nice to just read in this
+ * file from userland, this wouldn't work from a local zone.  So we have to have
+ * a zpool ioctl to return the complete configuration for all pools.  In the
+ * global zone, this will be identical to reading the file and unpacking it in
+ * userland.
+ */
+
+#include <errno.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <stddef.h>
+#include <string.h>
+#include <unistd.h>
+#include <libintl.h>
+#include <libuutil.h>
+
+#include "libzfs_impl.h"
+
+typedef struct config_node {
+	char		*cn_name;
+	nvlist_t	*cn_config;
+	uu_avl_node_t	cn_avl;
+} config_node_t;
+
+/* ARGSUSED */
+static int
+config_node_compare(const void *a, const void *b, void *unused)
+{
+	int ret;
+
+	const config_node_t *ca = (config_node_t *)a;
+	const config_node_t *cb = (config_node_t *)b;
+
+	ret = strcmp(ca->cn_name, cb->cn_name);
+
+	if (ret < 0)
+		return (-1);
+	else if (ret > 0)
+		return (1);
+	else
+		return (0);
+}
+
+void
+namespace_clear(libzfs_handle_t *hdl)
+{
+	if (hdl->libzfs_ns_avl) {
+		config_node_t *cn;
+		void *cookie = NULL;
+
+		while ((cn = uu_avl_teardown(hdl->libzfs_ns_avl,
+		    &cookie)) != NULL) {
+			nvlist_free(cn->cn_config);
+			free(cn->cn_name);
+			free(cn);
+		}
+
+		uu_avl_destroy(hdl->libzfs_ns_avl);
+		hdl->libzfs_ns_avl = NULL;
+	}
+
+	if (hdl->libzfs_ns_avlpool) {
+		uu_avl_pool_destroy(hdl->libzfs_ns_avlpool);
+		hdl->libzfs_ns_avlpool = NULL;
+	}
+}
+
+/*
+ * Loads the pool namespace, or re-loads it if the cache has changed.
+ */
+static int
+namespace_reload(libzfs_handle_t *hdl)
+{
+	nvlist_t *config;
+	config_node_t *cn;
+	nvpair_t *elem;
+	zfs_cmd_t zc = { 0 };
+	void *cookie;
+
+	if (hdl->libzfs_ns_gen == 0) {
+		/*
+		 * This is the first time we've accessed the configuration
+		 * cache.  Initialize the AVL tree and then fall through to the
+		 * common code.
+		 */
+		if ((hdl->libzfs_ns_avlpool = uu_avl_pool_create("config_pool",
+		    sizeof (config_node_t),
+		    offsetof(config_node_t, cn_avl),
+		    config_node_compare, UU_DEFAULT)) == NULL)
+			return (no_memory(hdl));
+
+		if ((hdl->libzfs_ns_avl = uu_avl_create(hdl->libzfs_ns_avlpool,
+		    NULL, UU_DEFAULT)) == NULL)
+			return (no_memory(hdl));
+	}
+
+	if (zcmd_alloc_dst_nvlist(hdl, &zc, 0) != 0)
+		return (-1);
+
+	for (;;) {
+		zc.zc_cookie = hdl->libzfs_ns_gen;
+		if (ioctl(hdl->libzfs_fd, ZFS_IOC_POOL_CONFIGS, &zc) != 0) {
+			switch (errno) {
+			case EEXIST:
+				/*
+				 * The namespace hasn't changed.
+				 */
+				zcmd_free_nvlists(&zc);
+				return (0);
+
+			case ENOMEM:
+				if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
+					zcmd_free_nvlists(&zc);
+					return (-1);
+				}
+				break;
+
+			default:
+				zcmd_free_nvlists(&zc);
+				return (zfs_standard_error(hdl, errno,
+				    dgettext(TEXT_DOMAIN, "failed to read "
+				    "pool configuration")));
+			}
+		} else {
+			hdl->libzfs_ns_gen = zc.zc_cookie;
+			break;
+		}
+	}
+
+	if (zcmd_read_dst_nvlist(hdl, &zc, &config) != 0) {
+		zcmd_free_nvlists(&zc);
+		return (-1);
+	}
+
+	zcmd_free_nvlists(&zc);
+
+	/*
+	 * Clear out any existing configuration information.
+	 */
+	cookie = NULL;
+	while ((cn = uu_avl_teardown(hdl->libzfs_ns_avl, &cookie)) != NULL) {
+		nvlist_free(cn->cn_config);
+		free(cn->cn_name);
+		free(cn);
+	}
+
+	elem = NULL;
+	while ((elem = nvlist_next_nvpair(config, elem)) != NULL) {
+		nvlist_t *child;
+		uu_avl_index_t where;
+
+		if ((cn = zfs_alloc(hdl, sizeof (config_node_t))) == NULL) {
+			nvlist_free(config);
+			return (-1);
+		}
+
+		if ((cn->cn_name = zfs_strdup(hdl,
+		    nvpair_name(elem))) == NULL) {
+			free(cn);
+			nvlist_free(config);
+			return (-1);
+		}
+
+		verify(nvpair_value_nvlist(elem, &child) == 0);
+		if (nvlist_dup(child, &cn->cn_config, 0) != 0) {
+			free(cn->cn_name);
+			free(cn);
+			nvlist_free(config);
+			return (no_memory(hdl));
+		}
+		verify(uu_avl_find(hdl->libzfs_ns_avl, cn, NULL, &where)
+		    == NULL);
+
+		uu_avl_insert(hdl->libzfs_ns_avl, cn, where);
+	}
+
+	nvlist_free(config);
+	return (0);
+}
+
+/*
+ * Retrieve the configuration for the given pool.  The configuration is a nvlist
+ * describing the vdevs, as well as the statistics associated with each one.
+ */
+nvlist_t *
+zpool_get_config(zpool_handle_t *zhp, nvlist_t **oldconfig)
+{
+	if (oldconfig)
+		*oldconfig = zhp->zpool_old_config;
+	return (zhp->zpool_config);
+}
+
+/*
+ * Retrieves a list of enabled features and their refcounts and caches it in
+ * the pool handle.
+ */
+nvlist_t *
+zpool_get_features(zpool_handle_t *zhp)
+{
+	nvlist_t *config, *features;
+
+	config = zpool_get_config(zhp, NULL);
+
+	if (config == NULL || !nvlist_exists(config,
+	    ZPOOL_CONFIG_FEATURE_STATS)) {
+		int error;
+		boolean_t missing = B_FALSE;
+
+		error = zpool_refresh_stats(zhp, &missing);
+
+		if (error != 0 || missing)
+			return (NULL);
+
+		config = zpool_get_config(zhp, NULL);
+	}
+
+	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURE_STATS,
+	    &features) == 0);
+
+	return (features);
+}
+
+/*
+ * Refresh the vdev statistics associated with the given pool.  This is used in
+ * iostat to show configuration changes and determine the delta from the last
+ * time the function was called.  This function can fail, in case the pool has
+ * been destroyed.
+ */
+int
+zpool_refresh_stats(zpool_handle_t *zhp, boolean_t *missing)
+{
+	zfs_cmd_t zc = { 0 };
+	int error;
+	nvlist_t *config;
+	libzfs_handle_t *hdl = zhp->zpool_hdl;
+
+	*missing = B_FALSE;
+	(void) strcpy(zc.zc_name, zhp->zpool_name);
+
+	if (zhp->zpool_config_size == 0)
+		zhp->zpool_config_size = 1 << 16;
+
+	if (zcmd_alloc_dst_nvlist(hdl, &zc, zhp->zpool_config_size) != 0)
+		return (-1);
+
+	for (;;) {
+		if (ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_POOL_STATS,
+		    &zc) == 0) {
+			/*
+			 * The real error is returned in the zc_cookie field.
+			 */
+			error = zc.zc_cookie;
+			break;
+		}
+
+		if (errno == ENOMEM) {
+			if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
+				zcmd_free_nvlists(&zc);
+				return (-1);
+			}
+		} else {
+			zcmd_free_nvlists(&zc);
+			if (errno == ENOENT || errno == EINVAL)
+				*missing = B_TRUE;
+			zhp->zpool_state = POOL_STATE_UNAVAIL;
+			return (0);
+		}
+	}
+
+	if (zcmd_read_dst_nvlist(hdl, &zc, &config) != 0) {
+		zcmd_free_nvlists(&zc);
+		return (-1);
+	}
+
+	zcmd_free_nvlists(&zc);
+
+	zhp->zpool_config_size = zc.zc_nvlist_dst_size;
+
+	if (zhp->zpool_config != NULL) {
+		uint64_t oldtxg, newtxg;
+
+		verify(nvlist_lookup_uint64(zhp->zpool_config,
+		    ZPOOL_CONFIG_POOL_TXG, &oldtxg) == 0);
+		verify(nvlist_lookup_uint64(config,
+		    ZPOOL_CONFIG_POOL_TXG, &newtxg) == 0);
+
+		if (zhp->zpool_old_config != NULL)
+			nvlist_free(zhp->zpool_old_config);
+
+		if (oldtxg != newtxg) {
+			nvlist_free(zhp->zpool_config);
+			zhp->zpool_old_config = NULL;
+		} else {
+			zhp->zpool_old_config = zhp->zpool_config;
+		}
+	}
+
+	zhp->zpool_config = config;
+	if (error)
+		zhp->zpool_state = POOL_STATE_UNAVAIL;
+	else
+		zhp->zpool_state = POOL_STATE_ACTIVE;
+
+	return (0);
+}
+
+/*
+ * Iterate over all pools in the system.
+ */
+int
+zpool_iter(libzfs_handle_t *hdl, zpool_iter_f func, void *data)
+{
+	config_node_t *cn;
+	zpool_handle_t *zhp;
+	int ret;
+
+	/*
+	 * If someone makes a recursive call to zpool_iter(), we want to avoid
+	 * refreshing the namespace because that will invalidate the parent
+	 * context.  We allow recursive calls, but simply re-use the same
+	 * namespace AVL tree.
+	 */
+	if (!hdl->libzfs_pool_iter && namespace_reload(hdl) != 0)
+		return (-1);
+
+	hdl->libzfs_pool_iter++;
+	for (cn = uu_avl_first(hdl->libzfs_ns_avl); cn != NULL;
+	    cn = uu_avl_next(hdl->libzfs_ns_avl, cn)) {
+
+		if (zpool_open_silent(hdl, cn->cn_name, &zhp) != 0) {
+			hdl->libzfs_pool_iter--;
+			return (-1);
+		}
+
+		if (zhp == NULL)
+			continue;
+
+		if ((ret = func(zhp, data)) != 0) {
+			hdl->libzfs_pool_iter--;
+			return (ret);
+		}
+	}
+	hdl->libzfs_pool_iter--;
+
+	return (0);
+}
+
+/*
+ * Iterate over root datasets, calling the given function for each.  The zfs
+ * handle passed each time must be explicitly closed by the callback.
+ */
+int
+zfs_iter_root(libzfs_handle_t *hdl, zfs_iter_f func, void *data)
+{
+	config_node_t *cn;
+	zfs_handle_t *zhp;
+	int ret;
+
+	if (namespace_reload(hdl) != 0)
+		return (-1);
+
+	for (cn = uu_avl_first(hdl->libzfs_ns_avl); cn != NULL;
+	    cn = uu_avl_next(hdl->libzfs_ns_avl, cn)) {
+
+		if ((zhp = make_dataset_handle(hdl, cn->cn_name)) == NULL)
+			continue;
+
+		if ((ret = func(zhp, data)) != 0)
+			return (ret);
+	}
+
+	return (0);
+}

lib/libzfs/common/libzfs_diff.c

@@ -0,0 +1,826 @@
+/*
+ * 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 (c) 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+/*
+ * zfs diff support
+ */
+#include <ctype.h>
+#include <errno.h>
+#include <libintl.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <attr.h>
+#include <stddef.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stropts.h>
+#include <pthread.h>
+#include <sys/zfs_ioctl.h>
+#include <libzfs.h>
+#include "libzfs_impl.h"
+
+#define	ZDIFF_SNAPDIR		"/.zfs/snapshot/"
+#define	ZDIFF_SHARESDIR 	"/.zfs/shares/"
+#define	ZDIFF_PREFIX		"zfs-diff-%d"
+
+#define	ZDIFF_ADDED	'+'
+#define	ZDIFF_MODIFIED	'M'
+#define	ZDIFF_REMOVED	'-'
+#define	ZDIFF_RENAMED	'R'
+
+static boolean_t
+do_name_cmp(const char *fpath, const char *tpath)
+{
+	char *fname, *tname;
+	fname = strrchr(fpath, '/') + 1;
+	tname = strrchr(tpath, '/') + 1;
+	return (strcmp(fname, tname) == 0);
+}
+
+typedef struct differ_info {
+	zfs_handle_t *zhp;
+	char *fromsnap;
+	char *frommnt;
+	char *tosnap;
+	char *tomnt;
+	char *ds;
+	char *dsmnt;
+	char *tmpsnap;
+	char errbuf[1024];
+	boolean_t isclone;
+	boolean_t scripted;
+	boolean_t classify;
+	boolean_t timestamped;
+	uint64_t shares;
+	int zerr;
+	int cleanupfd;
+	int outputfd;
+	int datafd;
+} differ_info_t;
+
+/*
+ * Given a {dsname, object id}, get the object path
+ */
+static int
+get_stats_for_obj(differ_info_t *di, const char *dsname, uint64_t obj,
+    char *pn, int maxlen, zfs_stat_t *sb)
+{
+	zfs_cmd_t zc = { 0 };
+	int error;
+
+	(void) strlcpy(zc.zc_name, dsname, sizeof (zc.zc_name));
+	zc.zc_obj = obj;
+
+	errno = 0;
+	error = ioctl(di->zhp->zfs_hdl->libzfs_fd, ZFS_IOC_OBJ_TO_STATS, &zc);
+	di->zerr = errno;
+
+	/* we can get stats even if we failed to get a path */
+	(void) memcpy(sb, &zc.zc_stat, sizeof (zfs_stat_t));
+	if (error == 0) {
+		ASSERT(di->zerr == 0);
+		(void) strlcpy(pn, zc.zc_value, maxlen);
+		return (0);
+	}
+
+	if (di->zerr == EPERM) {
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN,
+		    "The sys_config privilege or diff delegated permission "
+		    "is needed\nto discover path names"));
+		return (-1);
+	} else {
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN,
+		    "Unable to determine path or stats for "
+		    "object %lld in %s"), obj, dsname);
+		return (-1);
+	}
+}
+
+/*
+ * stream_bytes
+ *
+ * Prints a file name out a character at a time.  If the character is
+ * not in the range of what we consider "printable" ASCII, display it
+ * as an escaped 3-digit octal value.  ASCII values less than a space
+ * are all control characters and we declare the upper end as the
+ * DELete character.  This also is the last 7-bit ASCII character.
+ * We choose to treat all 8-bit ASCII as not printable for this
+ * application.
+ */
+static void
+stream_bytes(FILE *fp, const char *string)
+{
+	while (*string) {
+		if (*string > ' ' && *string != '\\' && *string < '\177')
+			(void) fprintf(fp, "%c", *string++);
+		else
+			(void) fprintf(fp, "\\%03o", *string++);
+	}
+}
+
+static void
+print_what(FILE *fp, mode_t what)
+{
+	char symbol;
+
+	switch (what & S_IFMT) {
+	case S_IFBLK:
+		symbol = 'B';
+		break;
+	case S_IFCHR:
+		symbol = 'C';
+		break;
+	case S_IFDIR:
+		symbol = '/';
+		break;
+	case S_IFDOOR:
+		symbol = '>';
+		break;
+	case S_IFIFO:
+		symbol = '|';
+		break;
+	case S_IFLNK:
+		symbol = '@';
+		break;
+	case S_IFPORT:
+		symbol = 'P';
+		break;
+	case S_IFSOCK:
+		symbol = '=';
+		break;
+	case S_IFREG:
+		symbol = 'F';
+		break;
+	default:
+		symbol = '?';
+		break;
+	}
+	(void) fprintf(fp, "%c", symbol);
+}
+
+static void
+print_cmn(FILE *fp, differ_info_t *di, const char *file)
+{
+	stream_bytes(fp, di->dsmnt);
+	stream_bytes(fp, file);
+}
+
+static void
+print_rename(FILE *fp, differ_info_t *di, const char *old, const char *new,
+    zfs_stat_t *isb)
+{
+	if (di->timestamped)
+		(void) fprintf(fp, "%10lld.%09lld\t",
+		    (longlong_t)isb->zs_ctime[0],
+		    (longlong_t)isb->zs_ctime[1]);
+	(void) fprintf(fp, "%c\t", ZDIFF_RENAMED);
+	if (di->classify) {
+		print_what(fp, isb->zs_mode);
+		(void) fprintf(fp, "\t");
+	}
+	print_cmn(fp, di, old);
+	if (di->scripted)
+		(void) fprintf(fp, "\t");
+	else
+		(void) fprintf(fp, " -> ");
+	print_cmn(fp, di, new);
+	(void) fprintf(fp, "\n");
+}
+
+static void
+print_link_change(FILE *fp, differ_info_t *di, int delta, const char *file,
+    zfs_stat_t *isb)
+{
+	if (di->timestamped)
+		(void) fprintf(fp, "%10lld.%09lld\t",
+		    (longlong_t)isb->zs_ctime[0],
+		    (longlong_t)isb->zs_ctime[1]);
+	(void) fprintf(fp, "%c\t", ZDIFF_MODIFIED);
+	if (di->classify) {
+		print_what(fp, isb->zs_mode);
+		(void) fprintf(fp, "\t");
+	}
+	print_cmn(fp, di, file);
+	(void) fprintf(fp, "\t(%+d)", delta);
+	(void) fprintf(fp, "\n");
+}
+
+static void
+print_file(FILE *fp, differ_info_t *di, char type, const char *file,
+    zfs_stat_t *isb)
+{
+	if (di->timestamped)
+		(void) fprintf(fp, "%10lld.%09lld\t",
+		    (longlong_t)isb->zs_ctime[0],
+		    (longlong_t)isb->zs_ctime[1]);
+	(void) fprintf(fp, "%c\t", type);
+	if (di->classify) {
+		print_what(fp, isb->zs_mode);
+		(void) fprintf(fp, "\t");
+	}
+	print_cmn(fp, di, file);
+	(void) fprintf(fp, "\n");
+}
+
+static int
+write_inuse_diffs_one(FILE *fp, differ_info_t *di, uint64_t dobj)
+{
+	struct zfs_stat fsb, tsb;
+	boolean_t same_name;
+	mode_t fmode, tmode;
+	char fobjname[MAXPATHLEN], tobjname[MAXPATHLEN];
+	int fobjerr, tobjerr;
+	int change;
+
+	if (dobj == di->shares)
+		return (0);
+
+	/*
+	 * Check the from and to snapshots for info on the object. If
+	 * we get ENOENT, then the object just didn't exist in that
+	 * snapshot.  If we get ENOTSUP, then we tried to get
+	 * info on a non-ZPL object, which we don't care about anyway.
+	 */
+	fobjerr = get_stats_for_obj(di, di->fromsnap, dobj, fobjname,
+	    MAXPATHLEN, &fsb);
+	if (fobjerr && di->zerr != ENOENT && di->zerr != ENOTSUP)
+		return (-1);
+
+	tobjerr = get_stats_for_obj(di, di->tosnap, dobj, tobjname,
+	    MAXPATHLEN, &tsb);
+	if (tobjerr && di->zerr != ENOENT && di->zerr != ENOTSUP)
+		return (-1);
+
+	/*
+	 * Unallocated object sharing the same meta dnode block
+	 */
+	if (fobjerr && tobjerr) {
+		ASSERT(di->zerr == ENOENT || di->zerr == ENOTSUP);
+		di->zerr = 0;
+		return (0);
+	}
+
+	di->zerr = 0; /* negate get_stats_for_obj() from side that failed */
+	fmode = fsb.zs_mode & S_IFMT;
+	tmode = tsb.zs_mode & S_IFMT;
+	if (fmode == S_IFDIR || tmode == S_IFDIR || fsb.zs_links == 0 ||
+	    tsb.zs_links == 0)
+		change = 0;
+	else
+		change = tsb.zs_links - fsb.zs_links;
+
+	if (fobjerr) {
+		if (change) {
+			print_link_change(fp, di, change, tobjname, &tsb);
+			return (0);
+		}
+		print_file(fp, di, ZDIFF_ADDED, tobjname, &tsb);
+		return (0);
+	} else if (tobjerr) {
+		if (change) {
+			print_link_change(fp, di, change, fobjname, &fsb);
+			return (0);
+		}
+		print_file(fp, di, ZDIFF_REMOVED, fobjname, &fsb);
+		return (0);
+	}
+
+	if (fmode != tmode && fsb.zs_gen == tsb.zs_gen)
+		tsb.zs_gen++;	/* Force a generational difference */
+	same_name = do_name_cmp(fobjname, tobjname);
+
+	/* Simple modification or no change */
+	if (fsb.zs_gen == tsb.zs_gen) {
+		/* No apparent changes.  Could we assert !this?  */
+		if (fsb.zs_ctime[0] == tsb.zs_ctime[0] &&
+		    fsb.zs_ctime[1] == tsb.zs_ctime[1])
+			return (0);
+		if (change) {
+			print_link_change(fp, di, change,
+			    change > 0 ? fobjname : tobjname, &tsb);
+		} else if (same_name) {
+			print_file(fp, di, ZDIFF_MODIFIED, fobjname, &tsb);
+		} else {
+			print_rename(fp, di, fobjname, tobjname, &tsb);
+		}
+		return (0);
+	} else {
+		/* file re-created or object re-used */
+		print_file(fp, di, ZDIFF_REMOVED, fobjname, &fsb);
+		print_file(fp, di, ZDIFF_ADDED, tobjname, &tsb);
+		return (0);
+	}
+}
+
+static int
+write_inuse_diffs(FILE *fp, differ_info_t *di, dmu_diff_record_t *dr)
+{
+	uint64_t o;
+	int err;
+
+	for (o = dr->ddr_first; o <= dr->ddr_last; o++) {
+		if (err = write_inuse_diffs_one(fp, di, o))
+			return (err);
+	}
+	return (0);
+}
+
+static int
+describe_free(FILE *fp, differ_info_t *di, uint64_t object, char *namebuf,
+    int maxlen)
+{
+	struct zfs_stat sb;
+
+	if (get_stats_for_obj(di, di->fromsnap, object, namebuf,
+	    maxlen, &sb) != 0) {
+		/* Let it slide, if in the delete queue on from side */
+		if (di->zerr == ENOENT && sb.zs_links == 0) {
+			di->zerr = 0;
+			return (0);
+		}
+		return (-1);
+	}
+
+	print_file(fp, di, ZDIFF_REMOVED, namebuf, &sb);
+	return (0);
+}
+
+static int
+write_free_diffs(FILE *fp, differ_info_t *di, dmu_diff_record_t *dr)
+{
+	zfs_cmd_t zc = { 0 };
+	libzfs_handle_t *lhdl = di->zhp->zfs_hdl;
+	char fobjname[MAXPATHLEN];
+
+	(void) strlcpy(zc.zc_name, di->fromsnap, sizeof (zc.zc_name));
+	zc.zc_obj = dr->ddr_first - 1;
+
+	ASSERT(di->zerr == 0);
+
+	while (zc.zc_obj < dr->ddr_last) {
+		int err;
+
+		err = ioctl(lhdl->libzfs_fd, ZFS_IOC_NEXT_OBJ, &zc);
+		if (err == 0) {
+			if (zc.zc_obj == di->shares) {
+				zc.zc_obj++;
+				continue;
+			}
+			if (zc.zc_obj > dr->ddr_last) {
+				break;
+			}
+			err = describe_free(fp, di, zc.zc_obj, fobjname,
+			    MAXPATHLEN);
+			if (err)
+				break;
+		} else if (errno == ESRCH) {
+			break;
+		} else {
+			(void) snprintf(di->errbuf, sizeof (di->errbuf),
+			    dgettext(TEXT_DOMAIN,
+			    "next allocated object (> %lld) find failure"),
+			    zc.zc_obj);
+			di->zerr = errno;
+			break;
+		}
+	}
+	if (di->zerr)
+		return (-1);
+	return (0);
+}
+
+static void *
+differ(void *arg)
+{
+	differ_info_t *di = arg;
+	dmu_diff_record_t dr;
+	FILE *ofp;
+	int err = 0;
+
+	if ((ofp = fdopen(di->outputfd, "w")) == NULL) {
+		di->zerr = errno;
+		(void) strerror_r(errno, di->errbuf, sizeof (di->errbuf));
+		(void) close(di->datafd);
+		return ((void *)-1);
+	}
+
+	for (;;) {
+		char *cp = (char *)&dr;
+		int len = sizeof (dr);
+		int rv;
+
+		do {
+			rv = read(di->datafd, cp, len);
+			cp += rv;
+			len -= rv;
+		} while (len > 0 && rv > 0);
+
+		if (rv < 0 || (rv == 0 && len != sizeof (dr))) {
+			di->zerr = EPIPE;
+			break;
+		} else if (rv == 0) {
+			/* end of file at a natural breaking point */
+			break;
+		}
+
+		switch (dr.ddr_type) {
+		case DDR_FREE:
+			err = write_free_diffs(ofp, di, &dr);
+			break;
+		case DDR_INUSE:
+			err = write_inuse_diffs(ofp, di, &dr);
+			break;
+		default:
+			di->zerr = EPIPE;
+			break;
+		}
+
+		if (err || di->zerr)
+			break;
+	}
+
+	(void) fclose(ofp);
+	(void) close(di->datafd);
+	if (err)
+		return ((void *)-1);
+	if (di->zerr) {
+		ASSERT(di->zerr == EINVAL);
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN,
+		    "Internal error: bad data from diff IOCTL"));
+		return ((void *)-1);
+	}
+	return ((void *)0);
+}
+
+static int
+find_shares_object(differ_info_t *di)
+{
+	char fullpath[MAXPATHLEN];
+	struct stat64 sb = { 0 };
+
+	(void) strlcpy(fullpath, di->dsmnt, MAXPATHLEN);
+	(void) strlcat(fullpath, ZDIFF_SHARESDIR, MAXPATHLEN);
+
+	if (stat64(fullpath, &sb) != 0) {
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN, "Cannot stat %s"), fullpath);
+		return (zfs_error(di->zhp->zfs_hdl, EZFS_DIFF, di->errbuf));
+	}
+
+	di->shares = (uint64_t)sb.st_ino;
+	return (0);
+}
+
+static int
+make_temp_snapshot(differ_info_t *di)
+{
+	libzfs_handle_t *hdl = di->zhp->zfs_hdl;
+	zfs_cmd_t zc = { 0 };
+
+	(void) snprintf(zc.zc_value, sizeof (zc.zc_value),
+	    ZDIFF_PREFIX, getpid());
+	(void) strlcpy(zc.zc_name, di->ds, sizeof (zc.zc_name));
+	zc.zc_cleanup_fd = di->cleanupfd;
+
+	if (ioctl(hdl->libzfs_fd, ZFS_IOC_TMP_SNAPSHOT, &zc) != 0) {
+		int err = errno;
+		if (err == EPERM) {
+			(void) snprintf(di->errbuf, sizeof (di->errbuf),
+			    dgettext(TEXT_DOMAIN, "The diff delegated "
+			    "permission is needed in order\nto create a "
+			    "just-in-time snapshot for diffing\n"));
+			return (zfs_error(hdl, EZFS_DIFF, di->errbuf));
+		} else {
+			(void) snprintf(di->errbuf, sizeof (di->errbuf),
+			    dgettext(TEXT_DOMAIN, "Cannot create just-in-time "
+			    "snapshot of '%s'"), zc.zc_name);
+			return (zfs_standard_error(hdl, err, di->errbuf));
+		}
+	}
+
+	di->tmpsnap = zfs_strdup(hdl, zc.zc_value);
+	di->tosnap = zfs_asprintf(hdl, "%s@%s", di->ds, di->tmpsnap);
+	return (0);
+}
+
+static void
+teardown_differ_info(differ_info_t *di)
+{
+	free(di->ds);
+	free(di->dsmnt);
+	free(di->fromsnap);
+	free(di->frommnt);
+	free(di->tosnap);
+	free(di->tmpsnap);
+	free(di->tomnt);
+	(void) close(di->cleanupfd);
+}
+
+static int
+get_snapshot_names(differ_info_t *di, const char *fromsnap,
+    const char *tosnap)
+{
+	libzfs_handle_t *hdl = di->zhp->zfs_hdl;
+	char *atptrf = NULL;
+	char *atptrt = NULL;
+	int fdslen, fsnlen;
+	int tdslen, tsnlen;
+
+	/*
+	 * Can accept
+	 *    dataset@snap1
+	 *    dataset@snap1 dataset@snap2
+	 *    dataset@snap1 @snap2
+	 *    dataset@snap1 dataset
+	 *    @snap1 dataset@snap2
+	 */
+	if (tosnap == NULL) {
+		/* only a from snapshot given, must be valid */
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN,
+		    "Badly formed snapshot name %s"), fromsnap);
+
+		if (!zfs_validate_name(hdl, fromsnap, ZFS_TYPE_SNAPSHOT,
+		    B_FALSE)) {
+			return (zfs_error(hdl, EZFS_INVALIDNAME,
+			    di->errbuf));
+		}
+
+		atptrf = strchr(fromsnap, '@');
+		ASSERT(atptrf != NULL);
+		fdslen = atptrf - fromsnap;
+
+		di->fromsnap = zfs_strdup(hdl, fromsnap);
+		di->ds = zfs_strdup(hdl, fromsnap);
+		di->ds[fdslen] = '\0';
+
+		/* the to snap will be a just-in-time snap of the head */
+		return (make_temp_snapshot(di));
+	}
+
+	(void) snprintf(di->errbuf, sizeof (di->errbuf),
+	    dgettext(TEXT_DOMAIN,
+	    "Unable to determine which snapshots to compare"));
+
+	atptrf = strchr(fromsnap, '@');
+	atptrt = strchr(tosnap, '@');
+	fdslen = atptrf ? atptrf - fromsnap : strlen(fromsnap);
+	tdslen = atptrt ? atptrt - tosnap : strlen(tosnap);
+	fsnlen = strlen(fromsnap) - fdslen;	/* includes @ sign */
+	tsnlen = strlen(tosnap) - tdslen;	/* includes @ sign */
+
+	if (fsnlen <= 1 || tsnlen == 1 || (fdslen == 0 && tdslen == 0) ||
+	    (fsnlen == 0 && tsnlen == 0)) {
+		return (zfs_error(hdl, EZFS_INVALIDNAME, di->errbuf));
+	} else if ((fdslen > 0 && tdslen > 0) &&
+	    ((tdslen != fdslen || strncmp(fromsnap, tosnap, fdslen) != 0))) {
+		/*
+		 * not the same dataset name, might be okay if
+		 * tosnap is a clone of a fromsnap descendant.
+		 */
+		char origin[ZFS_MAXNAMELEN];
+		zprop_source_t src;
+		zfs_handle_t *zhp;
+
+		di->ds = zfs_alloc(di->zhp->zfs_hdl, tdslen + 1);
+		(void) strncpy(di->ds, tosnap, tdslen);
+		di->ds[tdslen] = '\0';
+
+		zhp = zfs_open(hdl, di->ds, ZFS_TYPE_FILESYSTEM);
+		while (zhp != NULL) {
+			(void) zfs_prop_get(zhp, ZFS_PROP_ORIGIN,
+			    origin, sizeof (origin), &src, NULL, 0, B_FALSE);
+
+			if (strncmp(origin, fromsnap, fsnlen) == 0)
+				break;
+
+			(void) zfs_close(zhp);
+			zhp = zfs_open(hdl, origin, ZFS_TYPE_FILESYSTEM);
+		}
+
+		if (zhp == NULL) {
+			(void) snprintf(di->errbuf, sizeof (di->errbuf),
+			    dgettext(TEXT_DOMAIN,
+			    "Not an earlier snapshot from the same fs"));
+			return (zfs_error(hdl, EZFS_INVALIDNAME, di->errbuf));
+		} else {
+			(void) zfs_close(zhp);
+		}
+
+		di->isclone = B_TRUE;
+		di->fromsnap = zfs_strdup(hdl, fromsnap);
+		if (tsnlen) {
+			di->tosnap = zfs_strdup(hdl, tosnap);
+		} else {
+			return (make_temp_snapshot(di));
+		}
+	} else {
+		int dslen = fdslen ? fdslen : tdslen;
+
+		di->ds = zfs_alloc(hdl, dslen + 1);
+		(void) strncpy(di->ds, fdslen ? fromsnap : tosnap, dslen);
+		di->ds[dslen] = '\0';
+
+		di->fromsnap = zfs_asprintf(hdl, "%s%s", di->ds, atptrf);
+		if (tsnlen) {
+			di->tosnap = zfs_asprintf(hdl, "%s%s", di->ds, atptrt);
+		} else {
+			return (make_temp_snapshot(di));
+		}
+	}
+	return (0);
+}
+
+static int
+get_mountpoint(differ_info_t *di, char *dsnm, char **mntpt)
+{
+	boolean_t mounted;
+
+	mounted = is_mounted(di->zhp->zfs_hdl, dsnm, mntpt);
+	if (mounted == B_FALSE) {
+		(void) snprintf(di->errbuf, sizeof (di->errbuf),
+		    dgettext(TEXT_DOMAIN,
+		    "Cannot diff an unmounted snapshot"));
+		return (zfs_error(di->zhp->zfs_hdl, EZFS_BADTYPE, di->errbuf));
+	}
+
+	/* Avoid a double slash at the beginning of root-mounted datasets */
+	if (**mntpt == '/' && *(*mntpt + 1) == '\0')
+		**mntpt = '\0';
+	return (0);
+}
+
+static int
+get_mountpoints(differ_info_t *di)
+{
+	char *strptr;
+	char *frommntpt;
+
+	/*
+	 * first get the mountpoint for the parent dataset
+	 */
+	if (get_mountpoint(di, di->ds, &di->dsmnt) != 0)
+		return (-1);
+
+	strptr = strchr(di->tosnap, '@');
+	ASSERT3P(strptr, !=, NULL);
+	di->tomnt = zfs_asprintf(di->zhp->zfs_hdl, "%s%s%s", di->dsmnt,
+	    ZDIFF_SNAPDIR, ++strptr);
+
+	strptr = strchr(di->fromsnap, '@');
+	ASSERT3P(strptr, !=, NULL);
+
+	frommntpt = di->dsmnt;
+	if (di->isclone) {
+		char *mntpt;
+		int err;
+
+		*strptr = '\0';
+		err = get_mountpoint(di, di->fromsnap, &mntpt);
+		*strptr = '@';
+		if (err != 0)
+			return (-1);
+		frommntpt = mntpt;
+	}
+
+	di->frommnt = zfs_asprintf(di->zhp->zfs_hdl, "%s%s%s", frommntpt,
+	    ZDIFF_SNAPDIR, ++strptr);
+
+	if (di->isclone)
+		free(frommntpt);
+
+	return (0);
+}
+
+static int
+setup_differ_info(zfs_handle_t *zhp, const char *fromsnap,
+    const char *tosnap, differ_info_t *di)
+{
+	di->zhp = zhp;
+
+	di->cleanupfd = open(ZFS_DEV, O_RDWR|O_EXCL);
+	VERIFY(di->cleanupfd >= 0);
+
+	if (get_snapshot_names(di, fromsnap, tosnap) != 0)
+		return (-1);
+
+	if (get_mountpoints(di) != 0)
+		return (-1);
+
+	if (find_shares_object(di) != 0)
+		return (-1);
+
+	return (0);
+}
+
+int
+zfs_show_diffs(zfs_handle_t *zhp, int outfd, const char *fromsnap,
+    const char *tosnap, int flags)
+{
+	zfs_cmd_t zc = { 0 };
+	char errbuf[1024];
+	differ_info_t di = { 0 };
+	pthread_t tid;
+	int pipefd[2];
+	int iocerr;
+
+	(void) snprintf(errbuf, sizeof (errbuf),
+	    dgettext(TEXT_DOMAIN, "zfs diff failed"));
+
+	if (setup_differ_info(zhp, fromsnap, tosnap, &di)) {
+		teardown_differ_info(&di);
+		return (-1);
+	}
+
+	if (pipe(pipefd)) {
+		zfs_error_aux(zhp->zfs_hdl, strerror(errno));
+		teardown_differ_info(&di);
+		return (zfs_error(zhp->zfs_hdl, EZFS_PIPEFAILED, errbuf));
+	}
+
+	di.scripted = (flags & ZFS_DIFF_PARSEABLE);
+	di.classify = (flags & ZFS_DIFF_CLASSIFY);
+	di.timestamped = (flags & ZFS_DIFF_TIMESTAMP);
+
+	di.outputfd = outfd;
+	di.datafd = pipefd[0];
+
+	if (pthread_create(&tid, NULL, differ, &di)) {
+		zfs_error_aux(zhp->zfs_hdl, strerror(errno));
+		(void) close(pipefd[0]);
+		(void) close(pipefd[1]);
+		teardown_differ_info(&di);
+		return (zfs_error(zhp->zfs_hdl,
+		    EZFS_THREADCREATEFAILED, errbuf));
+	}
+
+	/* do the ioctl() */
+	(void) strlcpy(zc.zc_value, di.fromsnap, strlen(di.fromsnap) + 1);
+	(void) strlcpy(zc.zc_name, di.tosnap, strlen(di.tosnap) + 1);
+	zc.zc_cookie = pipefd[1];
+
+	iocerr = ioctl(zhp->zfs_hdl->libzfs_fd, ZFS_IOC_DIFF, &zc);
+	if (iocerr != 0) {
+		(void) snprintf(errbuf, sizeof (errbuf),
+		    dgettext(TEXT_DOMAIN, "Unable to obtain diffs"));
+		if (errno == EPERM) {
+			zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
+			    "\n   The sys_mount privilege or diff delegated "
+			    "permission is needed\n   to execute the "
+			    "diff ioctl"));
+		} else if (errno == EXDEV) {
+			zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
+			    "\n   Not an earlier snapshot from the same fs"));
+		} else if (errno != EPIPE || di.zerr == 0) {
+			zfs_error_aux(zhp->zfs_hdl, strerror(errno));
+		}
+		(void) close(pipefd[1]);
+		(void) pthread_cancel(tid);
+		(void) pthread_join(tid, NULL);
+		teardown_differ_info(&di);
+		if (di.zerr != 0 && di.zerr != EPIPE) {
+			zfs_error_aux(zhp->zfs_hdl, strerror(di.zerr));
+			return (zfs_error(zhp->zfs_hdl, EZFS_DIFF, di.errbuf));
+		} else {
+			return (zfs_error(zhp->zfs_hdl, EZFS_DIFFDATA, errbuf));
+		}
+	}
+
+	(void) close(pipefd[1]);
+	(void) pthread_join(tid, NULL);
+
+	if (di.zerr != 0) {
+		zfs_error_aux(zhp->zfs_hdl, strerror(di.zerr));
+		return (zfs_error(zhp->zfs_hdl, EZFS_DIFF, di.errbuf));
+	}
+	teardown_differ_info(&di);
+	return (0);
+}

lib/libzfs/common/libzfs_fru.c

@@ -0,0 +1,452 @@
+/*
+ * 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 2009 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <dlfcn.h>
+#include <errno.h>
+#include <libintl.h>
+#include <link.h>
+#include <pthread.h>
+#include <strings.h>
+#include <unistd.h>
+
+#include <libzfs.h>
+
+#include <fm/libtopo.h>
+#include <sys/fm/protocol.h>
+#include <sys/systeminfo.h>
+
+#include "libzfs_impl.h"
+
+/*
+ * This file is responsible for determining the relationship between I/O
+ * devices paths and physical locations.  In the world of MPxIO and external
+ * enclosures, the device path is not synonymous with the physical location.
+ * If you remove a drive and insert it into a different slot, it will end up
+ * with the same path under MPxIO.  If you recable storage enclosures, the
+ * device paths may change.  All of this makes it difficult to implement the
+ * 'autoreplace' property, which is supposed to automatically manage disk
+ * replacement based on physical slot.
+ *
+ * In order to work around these limitations, we have a per-vdev FRU property
+ * that is the libtopo path (minus disk-specific authority information) to the
+ * physical location of the device on the system.  This is an optional
+ * property, and is only needed when using the 'autoreplace' property or when
+ * generating FMA faults against vdevs.
+ */
+
+/*
+ * Because the FMA packages depend on ZFS, we have to dlopen() libtopo in case
+ * it is not present.  We only need this once per library instance, so it is
+ * not part of the libzfs handle.
+ */
+static void *_topo_dlhandle;
+static topo_hdl_t *(*_topo_open)(int, const char *, int *);
+static void (*_topo_close)(topo_hdl_t *);
+static char *(*_topo_snap_hold)(topo_hdl_t *, const char *, int *);
+static void (*_topo_snap_release)(topo_hdl_t *);
+static topo_walk_t *(*_topo_walk_init)(topo_hdl_t *, const char *,
+    topo_walk_cb_t, void *, int *);
+static int (*_topo_walk_step)(topo_walk_t *, int);
+static void (*_topo_walk_fini)(topo_walk_t *);
+static void (*_topo_hdl_strfree)(topo_hdl_t *, char *);
+static char *(*_topo_node_name)(tnode_t *);
+static int (*_topo_prop_get_string)(tnode_t *, const char *, const char *,
+    char **, int *);
+static int (*_topo_node_fru)(tnode_t *, nvlist_t **, nvlist_t *, int *);
+static int (*_topo_fmri_nvl2str)(topo_hdl_t *, nvlist_t *, char **, int *);
+static int (*_topo_fmri_strcmp_noauth)(topo_hdl_t *, const char *,
+    const char *);
+
+#define	ZFS_FRU_HASH_SIZE	257
+
+static size_t
+fru_strhash(const char *key)
+{
+	ulong_t g, h = 0;
+	const char *p;
+
+	for (p = key; *p != '\0'; p++) {
+		h = (h << 4) + *p;
+
+		if ((g = (h & 0xf0000000)) != 0) {
+			h ^= (g >> 24);
+			h ^= g;
+		}
+	}
+
+	return (h % ZFS_FRU_HASH_SIZE);
+}
+
+static int
+libzfs_fru_gather(topo_hdl_t *thp, tnode_t *tn, void *arg)
+{
+	libzfs_handle_t *hdl = arg;
+	nvlist_t *fru;
+	char *devpath, *frustr;
+	int err;
+	libzfs_fru_t *frup;
+	size_t idx;
+
+	/*
+	 * If this is the chassis node, and we don't yet have the system
+	 * chassis ID, then fill in this value now.
+	 */
+	if (hdl->libzfs_chassis_id[0] == '\0' &&
+	    strcmp(_topo_node_name(tn), "chassis") == 0) {
+		if (_topo_prop_get_string(tn, FM_FMRI_AUTHORITY,
+		    FM_FMRI_AUTH_CHASSIS, &devpath, &err) == 0)
+			(void) strlcpy(hdl->libzfs_chassis_id, devpath,
+			    sizeof (hdl->libzfs_chassis_id));
+	}
+
+	/*
+	 * Skip non-disk nodes.
+	 */
+	if (strcmp(_topo_node_name(tn), "disk") != 0)
+		return (TOPO_WALK_NEXT);
+
+	/*
+	 * Get the devfs path and FRU.
+	 */
+	if (_topo_prop_get_string(tn, "io", "devfs-path", &devpath, &err) != 0)
+		return (TOPO_WALK_NEXT);
+
+	if (libzfs_fru_lookup(hdl, devpath) != NULL) {
+		_topo_hdl_strfree(thp, devpath);
+		return (TOPO_WALK_NEXT);
+	}
+
+	if (_topo_node_fru(tn, &fru, NULL, &err) != 0) {
+		_topo_hdl_strfree(thp, devpath);
+		return (TOPO_WALK_NEXT);
+	}
+
+	/*
+	 * Convert the FRU into a string.
+	 */
+	if (_topo_fmri_nvl2str(thp, fru, &frustr, &err) != 0) {
+		nvlist_free(fru);
+		_topo_hdl_strfree(thp, devpath);
+		return (TOPO_WALK_NEXT);
+	}
+
+	nvlist_free(fru);
+
+	/*
+	 * Finally, we have a FRU string and device path.  Add it to the hash.
+	 */
+	if ((frup = calloc(sizeof (libzfs_fru_t), 1)) == NULL) {
+		_topo_hdl_strfree(thp, devpath);
+		_topo_hdl_strfree(thp, frustr);
+		return (TOPO_WALK_NEXT);
+	}
+
+	if ((frup->zf_device = strdup(devpath)) == NULL ||
+	    (frup->zf_fru = strdup(frustr)) == NULL) {
+		free(frup->zf_device);
+		free(frup);
+		_topo_hdl_strfree(thp, devpath);
+		_topo_hdl_strfree(thp, frustr);
+		return (TOPO_WALK_NEXT);
+	}
+
+	_topo_hdl_strfree(thp, devpath);
+	_topo_hdl_strfree(thp, frustr);
+
+	idx = fru_strhash(frup->zf_device);
+	frup->zf_chain = hdl->libzfs_fru_hash[idx];
+	hdl->libzfs_fru_hash[idx] = frup;
+	frup->zf_next = hdl->libzfs_fru_list;
+	hdl->libzfs_fru_list = frup;
+
+	return (TOPO_WALK_NEXT);
+}
+
+/*
+ * Called during initialization to setup the dynamic libtopo connection.
+ */
+#pragma init(libzfs_init_fru)
+static void
+libzfs_init_fru(void)
+{
+	char path[MAXPATHLEN];
+	char isa[257];
+
+#if defined(_LP64)
+	if (sysinfo(SI_ARCHITECTURE_64, isa, sizeof (isa)) < 0)
+		isa[0] = '\0';
+#else
+	isa[0] = '\0';
+#endif
+	(void) snprintf(path, sizeof (path),
+	    "/usr/lib/fm/%s/libtopo.so", isa);
+
+	if ((_topo_dlhandle = dlopen(path, RTLD_LAZY)) == NULL)
+		return;
+
+	_topo_open = (topo_hdl_t *(*)())
+	    dlsym(_topo_dlhandle, "topo_open");
+	_topo_close = (void (*)())
+	    dlsym(_topo_dlhandle, "topo_close");
+	_topo_snap_hold = (char *(*)())
+	    dlsym(_topo_dlhandle, "topo_snap_hold");
+	_topo_snap_release = (void (*)())
+	    dlsym(_topo_dlhandle, "topo_snap_release");
+	_topo_walk_init = (topo_walk_t *(*)())
+	    dlsym(_topo_dlhandle, "topo_walk_init");
+	_topo_walk_step = (int (*)())
+	    dlsym(_topo_dlhandle, "topo_walk_step");
+	_topo_walk_fini = (void (*)())
+	    dlsym(_topo_dlhandle, "topo_walk_fini");
+	_topo_hdl_strfree = (void (*)())
+	    dlsym(_topo_dlhandle, "topo_hdl_strfree");
+	_topo_node_name = (char *(*)())
+	    dlsym(_topo_dlhandle, "topo_node_name");
+	_topo_prop_get_string = (int (*)())
+	    dlsym(_topo_dlhandle, "topo_prop_get_string");
+	_topo_node_fru = (int (*)())
+	    dlsym(_topo_dlhandle, "topo_node_fru");
+	_topo_fmri_nvl2str = (int (*)())
+	    dlsym(_topo_dlhandle, "topo_fmri_nvl2str");
+	_topo_fmri_strcmp_noauth = (int (*)())
+	    dlsym(_topo_dlhandle, "topo_fmri_strcmp_noauth");
+
+	if (_topo_open == NULL || _topo_close == NULL ||
+	    _topo_snap_hold == NULL || _topo_snap_release == NULL ||
+	    _topo_walk_init == NULL || _topo_walk_step == NULL ||
+	    _topo_walk_fini == NULL || _topo_hdl_strfree == NULL ||
+	    _topo_node_name == NULL || _topo_prop_get_string == NULL ||
+	    _topo_node_fru == NULL || _topo_fmri_nvl2str == NULL ||
+	    _topo_fmri_strcmp_noauth == NULL) {
+		(void) dlclose(_topo_dlhandle);
+		_topo_dlhandle = NULL;
+	}
+}
+
+/*
+ * Refresh the mappings from device path -> FMRI.  We do this by walking the
+ * hc topology looking for disk nodes, and recording the io/devfs-path and FRU.
+ * Note that we strip out the disk-specific authority information (serial,
+ * part, revision, etc) so that we are left with only the identifying
+ * characteristics of the slot (hc path and chassis-id).
+ */
+void
+libzfs_fru_refresh(libzfs_handle_t *hdl)
+{
+	int err;
+	char *uuid;
+	topo_hdl_t *thp;
+	topo_walk_t *twp;
+
+	if (_topo_dlhandle == NULL)
+		return;
+
+	/*
+	 * Clear the FRU hash and initialize our basic structures.
+	 */
+	libzfs_fru_clear(hdl, B_FALSE);
+
+	if ((hdl->libzfs_topo_hdl = _topo_open(TOPO_VERSION,
+	    NULL, &err)) == NULL)
+		return;
+
+	thp = hdl->libzfs_topo_hdl;
+
+	if ((uuid = _topo_snap_hold(thp, NULL, &err)) == NULL)
+		return;
+
+	_topo_hdl_strfree(thp, uuid);
+
+	if (hdl->libzfs_fru_hash == NULL &&
+	    (hdl->libzfs_fru_hash =
+	    calloc(ZFS_FRU_HASH_SIZE * sizeof (void *), 1)) == NULL)
+		return;
+
+	/*
+	 * We now have a topo snapshot, so iterate over the hc topology looking
+	 * for disks to add to the hash.
+	 */
+	twp = _topo_walk_init(thp, FM_FMRI_SCHEME_HC,
+	    libzfs_fru_gather, hdl, &err);
+	if (twp != NULL) {
+		(void) _topo_walk_step(twp, TOPO_WALK_CHILD);
+		_topo_walk_fini(twp);
+	}
+}
+
+/*
+ * Given a devfs path, return the FRU for the device, if known.  This will
+ * automatically call libzfs_fru_refresh() if it hasn't already been called by
+ * the consumer.  The string returned is valid until the next call to
+ * libzfs_fru_refresh().
+ */
+const char *
+libzfs_fru_lookup(libzfs_handle_t *hdl, const char *devpath)
+{
+	size_t idx = fru_strhash(devpath);
+	libzfs_fru_t *frup;
+
+	if (hdl->libzfs_fru_hash == NULL)
+		libzfs_fru_refresh(hdl);
+
+	if (hdl->libzfs_fru_hash == NULL)
+		return (NULL);
+
+	for (frup = hdl->libzfs_fru_hash[idx]; frup != NULL;
+	    frup = frup->zf_chain) {
+		if (strcmp(devpath, frup->zf_device) == 0)
+			return (frup->zf_fru);
+	}
+
+	return (NULL);
+}
+
+/*
+ * Given a fru path, return the device path.  This will automatically call
+ * libzfs_fru_refresh() if it hasn't already been called by the consumer.  The
+ * string returned is valid until the next call to libzfs_fru_refresh().
+ */
+const char *
+libzfs_fru_devpath(libzfs_handle_t *hdl, const char *fru)
+{
+	libzfs_fru_t *frup;
+	size_t idx;
+
+	if (hdl->libzfs_fru_hash == NULL)
+		libzfs_fru_refresh(hdl);
+
+	if (hdl->libzfs_fru_hash == NULL)
+		return (NULL);
+
+	for (idx = 0; idx < ZFS_FRU_HASH_SIZE; idx++) {
+		for (frup = hdl->libzfs_fru_hash[idx]; frup != NULL;
+		    frup = frup->zf_next) {
+			if (_topo_fmri_strcmp_noauth(hdl->libzfs_topo_hdl,
+			    fru, frup->zf_fru))
+				return (frup->zf_device);
+		}
+	}
+
+	return (NULL);
+}
+
+/*
+ * Change the stored FRU for the given vdev.
+ */
+int
+zpool_fru_set(zpool_handle_t *zhp, uint64_t vdev_guid, const char *fru)
+{
+	zfs_cmd_t zc = { 0 };
+
+	(void) strncpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
+	(void) strncpy(zc.zc_value, fru, sizeof (zc.zc_value));
+	zc.zc_guid = vdev_guid;
+
+	if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_VDEV_SETFRU, &zc) != 0)
+		return (zpool_standard_error_fmt(zhp->zpool_hdl, errno,
+		    dgettext(TEXT_DOMAIN, "cannot set FRU")));
+
+	return (0);
+}
+
+/*
+ * Compare to two FRUs, ignoring any authority information.
+ */
+boolean_t
+libzfs_fru_compare(libzfs_handle_t *hdl, const char *a, const char *b)
+{
+	if (hdl->libzfs_fru_hash == NULL)
+		libzfs_fru_refresh(hdl);
+
+	if (hdl->libzfs_fru_hash == NULL)
+		return (strcmp(a, b) == 0);
+
+	return (_topo_fmri_strcmp_noauth(hdl->libzfs_topo_hdl, a, b));
+}
+
+/*
+ * This special function checks to see whether the FRU indicates it's supposed
+ * to be in the system chassis, but the chassis-id doesn't match.  This can
+ * happen in a clustered case, where both head nodes have the same logical
+ * disk, but opening the device on the other head node is meaningless.
+ */
+boolean_t
+libzfs_fru_notself(libzfs_handle_t *hdl, const char *fru)
+{
+	const char *chassisid;
+	size_t len;
+
+	if (hdl->libzfs_fru_hash == NULL)
+		libzfs_fru_refresh(hdl);
+
+	if (hdl->libzfs_chassis_id[0] == '\0')
+		return (B_FALSE);
+
+	if (strstr(fru, "/chassis=0/") == NULL)
+		return (B_FALSE);
+
+	if ((chassisid = strstr(fru, ":chassis-id=")) == NULL)
+		return (B_FALSE);
+
+	chassisid += 12;
+	len = strlen(hdl->libzfs_chassis_id);
+	if (strncmp(chassisid, hdl->libzfs_chassis_id, len) == 0 &&
+	    (chassisid[len] == '/' || chassisid[len] == ':'))
+		return (B_FALSE);
+
+	return (B_TRUE);
+}
+
+/*
+ * Clear memory associated with the FRU hash.
+ */
+void
+libzfs_fru_clear(libzfs_handle_t *hdl, boolean_t final)
+{
+	libzfs_fru_t *frup;
+
+	while ((frup = hdl->libzfs_fru_list) != NULL) {
+		hdl->libzfs_fru_list = frup->zf_next;
+		free(frup->zf_device);
+		free(frup->zf_fru);
+		free(frup);
+	}
+
+	hdl->libzfs_fru_list = NULL;
+
+	if (hdl->libzfs_topo_hdl != NULL) {
+		_topo_snap_release(hdl->libzfs_topo_hdl);
+		_topo_close(hdl->libzfs_topo_hdl);
+		hdl->libzfs_topo_hdl = NULL;
+	}
+
+	if (final) {
+		free(hdl->libzfs_fru_hash);
+	} else if (hdl->libzfs_fru_hash != NULL) {
+		bzero(hdl->libzfs_fru_hash,
+		    ZFS_FRU_HASH_SIZE * sizeof (void *));
+	}
+}

lib/libzfs/common/libzfs_impl.h

@@ -0,0 +1,216 @@
+/*
+ * CDDL HEADER SART
+ *
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
+ */
+
+#ifndef	_LIBFS_IMPL_H
+#define	_LIBFS_IMPL_H
+
+#include <sys/dmu.h>
+#include <sys/fs/zfs.h>
+#include <sys/zfs_ioctl.h>
+#include <sys/spa.h>
+#include <sys/nvpair.h>
+
+#include <libuutil.h>
+#include <libzfs.h>
+#include <libshare.h>
+
+#include <fm/libtopo.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+#ifdef	VERIFY
+#undef	VERIFY
+#endif
+#define	VERIFY	verify
+
+typedef struct libzfs_fru {
+	char *zf_device;
+	char *zf_fru;
+	struct libzfs_fru *zf_chain;
+	struct libzfs_fru *zf_next;
+} libzfs_fru_t;
+
+struct libzfs_handle {
+	int libzfs_error;
+	int libzfs_fd;
+	FILE *libzfs_mnttab;
+	FILE *libzfs_sharetab;
+	zpool_handle_t *libzfs_pool_handles;
+	uu_avl_pool_t *libzfs_ns_avlpool;
+	uu_avl_t *libzfs_ns_avl;
+	uint64_t libzfs_ns_gen;
+	int libzfs_desc_active;
+	char libzfs_action[1024];
+	char libzfs_desc[1024];
+	char *libzfs_log_str;
+	int libzfs_printerr;
+	int libzfs_storeerr; /* stuff error messages into buffer */
+	void *libzfs_sharehdl; /* libshare handle */
+	uint_t libzfs_shareflags;
+	boolean_t libzfs_mnttab_enable;
+	avl_tree_t libzfs_mnttab_cache;
+	int libzfs_pool_iter;
+	topo_hdl_t *libzfs_topo_hdl;
+	libzfs_fru_t **libzfs_fru_hash;
+	libzfs_fru_t *libzfs_fru_list;
+	char libzfs_chassis_id[256];
+};
+
+#define	ZFSSHARE_MISS	0x01	/* Didn't find entry in cache */
+
+struct zfs_handle {
+	libzfs_handle_t *zfs_hdl;
+	zpool_handle_t *zpool_hdl;
+	char zfs_name[ZFS_MAXNAMELEN];
+	zfs_type_t zfs_type; /* type including snapshot */
+	zfs_type_t zfs_head_type; /* type excluding snapshot */
+	dmu_objset_stats_t zfs_dmustats;
+	nvlist_t *zfs_props;
+	nvlist_t *zfs_user_props;
+	nvlist_t *zfs_recvd_props;
+	boolean_t zfs_mntcheck;
+	char *zfs_mntopts;
+	uint8_t *zfs_props_table;
+};
+
+/*
+ * This is different from checking zfs_type, because it will also catch
+ * snapshots of volumes.
+ */
+#define	ZFS_IS_VOLUME(zhp) ((zhp)->zfs_head_type == ZFS_TYPE_VOLUME)
+
+struct zpool_handle {
+	libzfs_handle_t *zpool_hdl;
+	zpool_handle_t *zpool_next;
+	char zpool_name[ZPOOL_MAXNAMELEN];
+	int zpool_state;
+	size_t zpool_config_size;
+	nvlist_t *zpool_config;
+	nvlist_t *zpool_old_config;
+	nvlist_t *zpool_props;
+	diskaddr_t zpool_start_block;
+};
+
+typedef enum {
+	PROTO_NFS = 0,
+	PROTO_SMB = 1,
+	PROTO_END = 2
+} zfs_share_proto_t;
+
+/*
+ * The following can be used as a bitmask and any new values
+ * added must preserve that capability.
+ */
+typedef enum {
+	SHARED_NOT_SHARED = 0x0,
+	SHARED_NFS = 0x2,
+	SHARED_SMB = 0x4
+} zfs_share_type_t;
+
+int zfs_error(libzfs_handle_t *, int, const char *);
+int zfs_error_fmt(libzfs_handle_t *, int, const char *, ...);
+void zfs_error_aux(libzfs_handle_t *, const char *, ...);
+void *zfs_alloc(libzfs_handle_t *, size_t);
+void *zfs_realloc(libzfs_handle_t *, void *, size_t, size_t);
+char *zfs_asprintf(libzfs_handle_t *, const char *, ...);
+char *zfs_strdup(libzfs_handle_t *, const char *);
+int no_memory(libzfs_handle_t *);
+
+int zfs_standard_error(libzfs_handle_t *, int, const char *);
+int zfs_standard_error_fmt(libzfs_handle_t *, int, const char *, ...);
+int zpool_standard_error(libzfs_handle_t *, int, const char *);
+int zpool_standard_error_fmt(libzfs_handle_t *, int, const char *, ...);
+
+int get_dependents(libzfs_handle_t *, boolean_t, const char *, char ***,
+    size_t *);
+zfs_handle_t *make_dataset_handle_zc(libzfs_handle_t *, zfs_cmd_t *);
+
+
+int zprop_parse_value(libzfs_handle_t *, nvpair_t *, int, zfs_type_t,
+    nvlist_t *, char **, uint64_t *, const char *);
+int zprop_expand_list(libzfs_handle_t *hdl, zprop_list_t **plp,
+    zfs_type_t type);
+
+/*
+ * Use this changelist_gather() flag to force attempting mounts
+ * on each change node regardless of whether or not it is currently
+ * mounted.
+ */
+#define	CL_GATHER_MOUNT_ALWAYS	1
+
+typedef struct prop_changelist prop_changelist_t;
+
+int zcmd_alloc_dst_nvlist(libzfs_handle_t *, zfs_cmd_t *, size_t);
+int zcmd_write_src_nvlist(libzfs_handle_t *, zfs_cmd_t *, nvlist_t *);
+int zcmd_write_conf_nvlist(libzfs_handle_t *, zfs_cmd_t *, nvlist_t *);
+int zcmd_expand_dst_nvlist(libzfs_handle_t *, zfs_cmd_t *);
+int zcmd_read_dst_nvlist(libzfs_handle_t *, zfs_cmd_t *, nvlist_t **);
+void zcmd_free_nvlists(zfs_cmd_t *);
+
+int changelist_prefix(prop_changelist_t *);
+int changelist_postfix(prop_changelist_t *);
+void changelist_rename(prop_changelist_t *, const char *, const char *);
+void changelist_remove(prop_changelist_t *, const char *);
+void changelist_free(prop_changelist_t *);
+prop_changelist_t *changelist_gather(zfs_handle_t *, zfs_prop_t, int, int);
+int changelist_unshare(prop_changelist_t *, zfs_share_proto_t *);
+int changelist_haszonedchild(prop_changelist_t *);
+
+void remove_mountpoint(zfs_handle_t *);
+int create_parents(libzfs_handle_t *, char *, int);
+boolean_t isa_child_of(const char *dataset, const char *parent);
+
+zfs_handle_t *make_dataset_handle(libzfs_handle_t *, const char *);
+
+int zpool_open_silent(libzfs_handle_t *, const char *, zpool_handle_t **);
+
+boolean_t zpool_name_valid(libzfs_handle_t *, boolean_t, const char *);
+
+int zfs_validate_name(libzfs_handle_t *hdl, const char *path, int type,
+    boolean_t modifying);
+
+void namespace_clear(libzfs_handle_t *);
+
+/*
+ * libshare (sharemgr) interfaces used internally.
+ */
+
+extern int zfs_init_libshare(libzfs_handle_t *, int);
+extern void zfs_uninit_libshare(libzfs_handle_t *);
+extern int zfs_parse_options(char *, zfs_share_proto_t);
+
+extern int zfs_unshare_proto(zfs_handle_t *,
+    const char *, zfs_share_proto_t *);
+
+extern void libzfs_fru_clear(libzfs_handle_t *, boolean_t);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _LIBFS_IMPL_H */

lib/libzfs/common/libzfs_iter.c

@@ -0,0 +1,462 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright 2010 Nexenta Systems, Inc. All rights reserved.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <strings.h>
+#include <unistd.h>
+#include <stddef.h>
+#include <libintl.h>
+#include <libzfs.h>
+
+#include "libzfs_impl.h"
+
+int
+zfs_iter_clones(zfs_handle_t *zhp, zfs_iter_f func, void *data)
+{
+	nvlist_t *nvl = zfs_get_clones_nvl(zhp);
+	nvpair_t *pair;
+
+	if (nvl == NULL)
+		return (0);
+
+	for (pair = nvlist_next_nvpair(nvl, NULL); pair != NULL;
+	    pair = nvlist_next_nvpair(nvl, pair)) {
+		zfs_handle_t *clone = zfs_open(zhp->zfs_hdl, nvpair_name(pair),
+		    ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME);
+		if (clone != NULL) {
+			int err = func(clone, data);
+			if (err != 0)
+				return (err);
+		}
+	}
+	return (0);
+}
+
+static int
+zfs_do_list_ioctl(zfs_handle_t *zhp, int arg, zfs_cmd_t *zc)
+{
+	int rc;
+	uint64_t	orig_cookie;
+
+	orig_cookie = zc->zc_cookie;
+top:
+	(void) strlcpy(zc->zc_name, zhp->zfs_name, sizeof (zc->zc_name));
+	rc = ioctl(zhp->zfs_hdl->libzfs_fd, arg, zc);
+
+	if (rc == -1) {
+		switch (errno) {
+		case ENOMEM:
+			/* expand nvlist memory and try again */
+			if (zcmd_expand_dst_nvlist(zhp->zfs_hdl, zc) != 0) {
+				zcmd_free_nvlists(zc);
+				return (-1);
+			}
+			zc->zc_cookie = orig_cookie;
+			goto top;
+		/*
+		 * An errno value of ESRCH indicates normal completion.
+		 * If ENOENT is returned, then the underlying dataset
+		 * has been removed since we obtained the handle.
+		 */
+		case ESRCH:
+		case ENOENT:
+			rc = 1;
+			break;
+		default:
+			rc = zfs_standard_error(zhp->zfs_hdl, errno,
+			    dgettext(TEXT_DOMAIN,
+			    "cannot iterate filesystems"));
+			break;
+		}
+	}
+	return (rc);
+}
+
+/*
+ * Iterate over all child filesystems
+ */
+int
+zfs_iter_filesystems(zfs_handle_t *zhp, zfs_iter_f func, void *data)
+{
+	zfs_cmd_t zc = { 0 };
+	zfs_handle_t *nzhp;
+	int ret;
+
+	if (zhp->zfs_type != ZFS_TYPE_FILESYSTEM)
+		return (0);
+
+	if (zcmd_alloc_dst_nvlist(zhp->zfs_hdl, &zc, 0) != 0)
+		return (-1);
+
+	while ((ret = zfs_do_list_ioctl(zhp, ZFS_IOC_DATASET_LIST_NEXT,
+	    &zc)) == 0) {
+		/*
+		 * Silently ignore errors, as the only plausible explanation is
+		 * that the pool has since been removed.
+		 */
+		if ((nzhp = make_dataset_handle_zc(zhp->zfs_hdl,
+		    &zc)) == NULL) {
+			continue;
+		}
+
+		if ((ret = func(nzhp, data)) != 0) {
+			zcmd_free_nvlists(&zc);
+			return (ret);
+		}
+	}
+	zcmd_free_nvlists(&zc);
+	return ((ret < 0) ? ret : 0);
+}
+
+/*
+ * Iterate over all snapshots
+ */
+int
+zfs_iter_snapshots(zfs_handle_t *zhp, zfs_iter_f func, void *data)
+{
+	zfs_cmd_t zc = { 0 };
+	zfs_handle_t *nzhp;
+	int ret;
+
+	if (zhp->zfs_type == ZFS_TYPE_SNAPSHOT)
+		return (0);
+
+	if (zcmd_alloc_dst_nvlist(zhp->zfs_hdl, &zc, 0) != 0)
+		return (-1);
+	while ((ret = zfs_do_list_ioctl(zhp, ZFS_IOC_SNAPSHOT_LIST_NEXT,
+	    &zc)) == 0) {
+
+		if ((nzhp = make_dataset_handle_zc(zhp->zfs_hdl,
+		    &zc)) == NULL) {
+			continue;
+		}
+
+		if ((ret = func(nzhp, data)) != 0) {
+			zcmd_free_nvlists(&zc);
+			return (ret);
+		}
+	}
+	zcmd_free_nvlists(&zc);
+	return ((ret < 0) ? ret : 0);
+}
+
+/*
+ * Routines for dealing with the sorted snapshot functionality
+ */
+typedef struct zfs_node {
+	zfs_handle_t	*zn_handle;
+	avl_node_t	zn_avlnode;
+} zfs_node_t;
+
+static int
+zfs_sort_snaps(zfs_handle_t *zhp, void *data)
+{
+	avl_tree_t *avl = data;
+	zfs_node_t *node;
+	zfs_node_t search;
+
+	search.zn_handle = zhp;
+	node = avl_find(avl, &search, NULL);
+	if (node) {
+		/*
+		 * If this snapshot was renamed while we were creating the
+		 * AVL tree, it's possible that we already inserted it under
+		 * its old name. Remove the old handle before adding the new
+		 * one.
+		 */
+		zfs_close(node->zn_handle);
+		avl_remove(avl, node);
+		free(node);
+	}
+
+	node = zfs_alloc(zhp->zfs_hdl, sizeof (zfs_node_t));
+	node->zn_handle = zhp;
+	avl_add(avl, node);
+
+	return (0);
+}
+
+static int
+zfs_snapshot_compare(const void *larg, const void *rarg)
+{
+	zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle;
+	zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle;
+	uint64_t lcreate, rcreate;
+
+	/*
+	 * Sort them according to creation time.  We use the hidden
+	 * CREATETXG property to get an absolute ordering of snapshots.
+	 */
+	lcreate = zfs_prop_get_int(l, ZFS_PROP_CREATETXG);
+	rcreate = zfs_prop_get_int(r, ZFS_PROP_CREATETXG);
+
+	if (lcreate < rcreate)
+		return (-1);
+	else if (lcreate > rcreate)
+		return (+1);
+	else
+		return (0);
+}
+
+int
+zfs_iter_snapshots_sorted(zfs_handle_t *zhp, zfs_iter_f callback, void *data)
+{
+	int ret = 0;
+	zfs_node_t *node;
+	avl_tree_t avl;
+	void *cookie = NULL;
+
+	avl_create(&avl, zfs_snapshot_compare,
+	    sizeof (zfs_node_t), offsetof(zfs_node_t, zn_avlnode));
+
+	ret = zfs_iter_snapshots(zhp, zfs_sort_snaps, &avl);
+
+	for (node = avl_first(&avl); node != NULL; node = AVL_NEXT(&avl, node))
+		ret |= callback(node->zn_handle, data);
+
+	while ((node = avl_destroy_nodes(&avl, &cookie)) != NULL)
+		free(node);
+
+	avl_destroy(&avl);
+
+	return (ret);
+}
+
+typedef struct {
+	char *ssa_first;
+	char *ssa_last;
+	boolean_t ssa_seenfirst;
+	boolean_t ssa_seenlast;
+	zfs_iter_f ssa_func;
+	void *ssa_arg;
+} snapspec_arg_t;
+
+static int
+snapspec_cb(zfs_handle_t *zhp, void *arg) {
+	snapspec_arg_t *ssa = arg;
+	char *shortsnapname;
+	int err = 0;
+
+	if (ssa->ssa_seenlast)
+		return (0);
+	shortsnapname = zfs_strdup(zhp->zfs_hdl,
+	    strchr(zfs_get_name(zhp), '@') + 1);
+
+	if (!ssa->ssa_seenfirst && strcmp(shortsnapname, ssa->ssa_first) == 0)
+		ssa->ssa_seenfirst = B_TRUE;
+
+	if (ssa->ssa_seenfirst) {
+		err = ssa->ssa_func(zhp, ssa->ssa_arg);
+	} else {
+		zfs_close(zhp);
+	}
+
+	if (strcmp(shortsnapname, ssa->ssa_last) == 0)
+		ssa->ssa_seenlast = B_TRUE;
+	free(shortsnapname);
+
+	return (err);
+}
+
+/*
+ * spec is a string like "A,B%C,D"
+ *
+ * <snaps>, where <snaps> can be:
+ *      <snap>          (single snapshot)
+ *      <snap>%<snap>   (range of snapshots, inclusive)
+ *      %<snap>         (range of snapshots, starting with earliest)
+ *      <snap>%         (range of snapshots, ending with last)
+ *      %               (all snapshots)
+ *      <snaps>[,...]   (comma separated list of the above)
+ *
+ * If a snapshot can not be opened, continue trying to open the others, but
+ * return ENOENT at the end.
+ */
+int
+zfs_iter_snapspec(zfs_handle_t *fs_zhp, const char *spec_orig,
+    zfs_iter_f func, void *arg)
+{
+	char buf[ZFS_MAXNAMELEN];
+	char *comma_separated, *cp;
+	int err = 0;
+	int ret = 0;
+
+	(void) strlcpy(buf, spec_orig, sizeof (buf));
+	cp = buf;
+
+	while ((comma_separated = strsep(&cp, ",")) != NULL) {
+		char *pct = strchr(comma_separated, '%');
+		if (pct != NULL) {
+			snapspec_arg_t ssa = { 0 };
+			ssa.ssa_func = func;
+			ssa.ssa_arg = arg;
+
+			if (pct == comma_separated)
+				ssa.ssa_seenfirst = B_TRUE;
+			else
+				ssa.ssa_first = comma_separated;
+			*pct = '\0';
+			ssa.ssa_last = pct + 1;
+
+			/*
+			 * If there is a lastname specified, make sure it
+			 * exists.
+			 */
+			if (ssa.ssa_last[0] != '\0') {
+				char snapname[ZFS_MAXNAMELEN];
+				(void) snprintf(snapname, sizeof (snapname),
+				    "%s@%s", zfs_get_name(fs_zhp),
+				    ssa.ssa_last);
+				if (!zfs_dataset_exists(fs_zhp->zfs_hdl,
+				    snapname, ZFS_TYPE_SNAPSHOT)) {
+					ret = ENOENT;
+					continue;
+				}
+			}
+
+			err = zfs_iter_snapshots_sorted(fs_zhp,
+			    snapspec_cb, &ssa);
+			if (ret == 0)
+				ret = err;
+			if (ret == 0 && (!ssa.ssa_seenfirst ||
+			    (ssa.ssa_last[0] != '\0' && !ssa.ssa_seenlast))) {
+				ret = ENOENT;
+			}
+		} else {
+			char snapname[ZFS_MAXNAMELEN];
+			zfs_handle_t *snap_zhp;
+			(void) snprintf(snapname, sizeof (snapname), "%s@%s",
+			    zfs_get_name(fs_zhp), comma_separated);
+			snap_zhp = make_dataset_handle(fs_zhp->zfs_hdl,
+			    snapname);
+			if (snap_zhp == NULL) {
+				ret = ENOENT;
+				continue;
+			}
+			err = func(snap_zhp, arg);
+			if (ret == 0)
+				ret = err;
+		}
+	}
+
+	return (ret);
+}
+
+/*
+ * Iterate over all children, snapshots and filesystems
+ */
+int
+zfs_iter_children(zfs_handle_t *zhp, zfs_iter_f func, void *data)
+{
+	int ret;
+
+	if ((ret = zfs_iter_filesystems(zhp, func, data)) != 0)
+		return (ret);
+
+	return (zfs_iter_snapshots(zhp, func, data));
+}
+
+
+typedef struct iter_stack_frame {
+	struct iter_stack_frame *next;
+	zfs_handle_t *zhp;
+} iter_stack_frame_t;
+
+typedef struct iter_dependents_arg {
+	boolean_t first;
+	boolean_t allowrecursion;
+	iter_stack_frame_t *stack;
+	zfs_iter_f func;
+	void *data;
+} iter_dependents_arg_t;
+
+static int
+iter_dependents_cb(zfs_handle_t *zhp, void *arg)
+{
+	iter_dependents_arg_t *ida = arg;
+	int err;
+	boolean_t first = ida->first;
+	ida->first = B_FALSE;
+
+	if (zhp->zfs_type == ZFS_TYPE_SNAPSHOT) {
+		err = zfs_iter_clones(zhp, iter_dependents_cb, ida);
+	} else {
+		iter_stack_frame_t isf;
+		iter_stack_frame_t *f;
+
+		/*
+		 * check if there is a cycle by seeing if this fs is already
+		 * on the stack.
+		 */
+		for (f = ida->stack; f != NULL; f = f->next) {
+			if (f->zhp->zfs_dmustats.dds_guid ==
+			    zhp->zfs_dmustats.dds_guid) {
+				if (ida->allowrecursion) {
+					zfs_close(zhp);
+					return (0);
+				} else {
+					zfs_error_aux(zhp->zfs_hdl,
+					    dgettext(TEXT_DOMAIN,
+					    "recursive dependency at '%s'"),
+					    zfs_get_name(zhp));
+					err = zfs_error(zhp->zfs_hdl,
+					    EZFS_RECURSIVE,
+					    dgettext(TEXT_DOMAIN,
+					    "cannot determine dependent "
+					    "datasets"));
+					zfs_close(zhp);
+					return (err);
+				}
+			}
+		}
+
+		isf.zhp = zhp;
+		isf.next = ida->stack;
+		ida->stack = &isf;
+		err = zfs_iter_filesystems(zhp, iter_dependents_cb, ida);
+		if (err == 0)
+			err = zfs_iter_snapshots(zhp, iter_dependents_cb, ida);
+		ida->stack = isf.next;
+	}
+	if (!first && err == 0)
+		err = ida->func(zhp, ida->data);
+	return (err);
+}
+
+int
+zfs_iter_dependents(zfs_handle_t *zhp, boolean_t allowrecursion,
+    zfs_iter_f func, void *data)
+{
+	iter_dependents_arg_t ida;
+	ida.allowrecursion = allowrecursion;
+	ida.stack = NULL;
+	ida.func = func;
+	ida.data = data;
+	ida.first = B_TRUE;
+	return (iter_dependents_cb(zfs_handle_dup(zhp), &ida));
+}

lib/libzfs/common/libzfs_status.c

@@ -0,0 +1,414 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2012 by Delphix. All rights reserved.
+ */
+
+/*
+ * This file contains the functions which analyze the status of a pool.  This
+ * include both the status of an active pool, as well as the status exported
+ * pools.  Returns one of the ZPOOL_STATUS_* defines describing the status of
+ * the pool.  This status is independent (to a certain degree) from the state of
+ * the pool.  A pool's state describes only whether or not it is capable of
+ * providing the necessary fault tolerance for data.  The status describes the
+ * overall status of devices.  A pool that is online can still have a device
+ * that is experiencing errors.
+ *
+ * Only a subset of the possible faults can be detected using 'zpool status',
+ * and not all possible errors correspond to a FMA message ID.  The explanation
+ * is left up to the caller, depending on whether it is a live pool or an
+ * import.
+ */
+
+#include <libzfs.h>
+#include <string.h>
+#include <unistd.h>
+#include "libzfs_impl.h"
+
+/*
+ * Message ID table.  This must be kept in sync with the ZPOOL_STATUS_* defines
+ * in libzfs.h.  Note that there are some status results which go past the end
+ * of this table, and hence have no associated message ID.
+ */
+static char *zfs_msgid_table[] = {
+	"ZFS-8000-14",
+	"ZFS-8000-2Q",
+	"ZFS-8000-3C",
+	"ZFS-8000-4J",
+	"ZFS-8000-5E",
+	"ZFS-8000-6X",
+	"ZFS-8000-72",
+	"ZFS-8000-8A",
+	"ZFS-8000-9P",
+	"ZFS-8000-A5",
+	"ZFS-8000-EY",
+	"ZFS-8000-HC",
+	"ZFS-8000-JQ",
+	"ZFS-8000-K4",
+};
+
+#define	NMSGID	(sizeof (zfs_msgid_table) / sizeof (zfs_msgid_table[0]))
+
+/* ARGSUSED */
+static int
+vdev_missing(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_CANT_OPEN &&
+	    aux == VDEV_AUX_OPEN_FAILED);
+}
+
+/* ARGSUSED */
+static int
+vdev_faulted(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_FAULTED);
+}
+
+/* ARGSUSED */
+static int
+vdev_errors(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_DEGRADED || errs != 0);
+}
+
+/* ARGSUSED */
+static int
+vdev_broken(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_CANT_OPEN);
+}
+
+/* ARGSUSED */
+static int
+vdev_offlined(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_OFFLINE);
+}
+
+/* ARGSUSED */
+static int
+vdev_removed(uint64_t state, uint64_t aux, uint64_t errs)
+{
+	return (state == VDEV_STATE_REMOVED);
+}
+
+/*
+ * Detect if any leaf devices that have seen errors or could not be opened.
+ */
+static boolean_t
+find_vdev_problem(nvlist_t *vdev, int (*func)(uint64_t, uint64_t, uint64_t))
+{
+	nvlist_t **child;
+	vdev_stat_t *vs;
+	uint_t c, children;
+	char *type;
+
+	/*
+	 * Ignore problems within a 'replacing' vdev, since we're presumably in
+	 * the process of repairing any such errors, and don't want to call them
+	 * out again.  We'll pick up the fact that a resilver is happening
+	 * later.
+	 */
+	verify(nvlist_lookup_string(vdev, ZPOOL_CONFIG_TYPE, &type) == 0);
+	if (strcmp(type, VDEV_TYPE_REPLACING) == 0)
+		return (B_FALSE);
+
+	if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, &child,
+	    &children) == 0) {
+		for (c = 0; c < children; c++)
+			if (find_vdev_problem(child[c], func))
+				return (B_TRUE);
+	} else {
+		verify(nvlist_lookup_uint64_array(vdev, ZPOOL_CONFIG_VDEV_STATS,
+		    (uint64_t **)&vs, &c) == 0);
+
+		if (func(vs->vs_state, vs->vs_aux,
+		    vs->vs_read_errors +
+		    vs->vs_write_errors +
+		    vs->vs_checksum_errors))
+			return (B_TRUE);
+	}
+
+	return (B_FALSE);
+}
+
+/*
+ * Active pool health status.
+ *
+ * To determine the status for a pool, we make several passes over the config,
+ * picking the most egregious error we find.  In order of importance, we do the
+ * following:
+ *
+ *	- Check for a complete and valid configuration
+ *	- Look for any faulted or missing devices in a non-replicated config
+ *	- Check for any data errors
+ *	- Check for any faulted or missing devices in a replicated config
+ *	- Look for any devices showing errors
+ *	- Check for any resilvering devices
+ *
+ * There can obviously be multiple errors within a single pool, so this routine
+ * only picks the most damaging of all the current errors to report.
+ */
+static zpool_status_t
+check_status(nvlist_t *config, boolean_t isimport)
+{
+	nvlist_t *nvroot;
+	vdev_stat_t *vs;
+	pool_scan_stat_t *ps = NULL;
+	uint_t vsc, psc;
+	uint64_t nerr;
+	uint64_t version;
+	uint64_t stateval;
+	uint64_t suspended;
+	uint64_t hostid = 0;
+
+	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
+	    &version) == 0);
+	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
+	    &nvroot) == 0);
+	verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS,
+	    (uint64_t **)&vs, &vsc) == 0);
+	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
+	    &stateval) == 0);
+
+	/*
+	 * Currently resilvering a vdev
+	 */
+	(void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS,
+	    (uint64_t **)&ps, &psc);
+	if (ps && ps->pss_func == POOL_SCAN_RESILVER &&
+	    ps->pss_state == DSS_SCANNING)
+		return (ZPOOL_STATUS_RESILVERING);
+
+	/*
+	 * Pool last accessed by another system.
+	 */
+	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid);
+	if (hostid != 0 && (unsigned long)hostid != gethostid() &&
+	    stateval == POOL_STATE_ACTIVE)
+		return (ZPOOL_STATUS_HOSTID_MISMATCH);
+
+	/*
+	 * Newer on-disk version.
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    vs->vs_aux == VDEV_AUX_VERSION_NEWER)
+		return (ZPOOL_STATUS_VERSION_NEWER);
+
+	/*
+	 * Unsupported feature(s).
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    vs->vs_aux == VDEV_AUX_UNSUP_FEAT) {
+		nvlist_t *nvinfo;
+
+		verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO,
+		    &nvinfo) == 0);
+		if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY))
+			return (ZPOOL_STATUS_UNSUP_FEAT_WRITE);
+		return (ZPOOL_STATUS_UNSUP_FEAT_READ);
+	}
+
+	/*
+	 * Check that the config is complete.
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    vs->vs_aux == VDEV_AUX_BAD_GUID_SUM)
+		return (ZPOOL_STATUS_BAD_GUID_SUM);
+
+	/*
+	 * Check whether the pool has suspended due to failed I/O.
+	 */
+	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED,
+	    &suspended) == 0) {
+		if (suspended == ZIO_FAILURE_MODE_CONTINUE)
+			return (ZPOOL_STATUS_IO_FAILURE_CONTINUE);
+		return (ZPOOL_STATUS_IO_FAILURE_WAIT);
+	}
+
+	/*
+	 * Could not read a log.
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    vs->vs_aux == VDEV_AUX_BAD_LOG) {
+		return (ZPOOL_STATUS_BAD_LOG);
+	}
+
+	/*
+	 * Bad devices in non-replicated config.
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    find_vdev_problem(nvroot, vdev_faulted))
+		return (ZPOOL_STATUS_FAULTED_DEV_NR);
+
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    find_vdev_problem(nvroot, vdev_missing))
+		return (ZPOOL_STATUS_MISSING_DEV_NR);
+
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    find_vdev_problem(nvroot, vdev_broken))
+		return (ZPOOL_STATUS_CORRUPT_LABEL_NR);
+
+	/*
+	 * Corrupted pool metadata
+	 */
+	if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
+	    vs->vs_aux == VDEV_AUX_CORRUPT_DATA)
+		return (ZPOOL_STATUS_CORRUPT_POOL);
+
+	/*
+	 * Persistent data errors.
+	 */
+	if (!isimport) {
+		if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT,
+		    &nerr) == 0 && nerr != 0)
+			return (ZPOOL_STATUS_CORRUPT_DATA);
+	}
+
+	/*
+	 * Missing devices in a replicated config.
+	 */
+	if (find_vdev_problem(nvroot, vdev_faulted))
+		return (ZPOOL_STATUS_FAULTED_DEV_R);
+	if (find_vdev_problem(nvroot, vdev_missing))
+		return (ZPOOL_STATUS_MISSING_DEV_R);
+	if (find_vdev_problem(nvroot, vdev_broken))
+		return (ZPOOL_STATUS_CORRUPT_LABEL_R);
+
+	/*
+	 * Devices with errors
+	 */
+	if (!isimport && find_vdev_problem(nvroot, vdev_errors))
+		return (ZPOOL_STATUS_FAILING_DEV);
+
+	/*
+	 * Offlined devices
+	 */
+	if (find_vdev_problem(nvroot, vdev_offlined))
+		return (ZPOOL_STATUS_OFFLINE_DEV);
+
+	/*
+	 * Removed device
+	 */
+	if (find_vdev_problem(nvroot, vdev_removed))
+		return (ZPOOL_STATUS_REMOVED_DEV);
+
+	/*
+	 * Outdated, but usable, version
+	 */
+	if (SPA_VERSION_IS_SUPPORTED(version) && version != SPA_VERSION)
+		return (ZPOOL_STATUS_VERSION_OLDER);
+
+	return (ZPOOL_STATUS_OK);
+}
+
+zpool_status_t
+zpool_get_status(zpool_handle_t *zhp, char **msgid)
+{
+	zpool_status_t ret = check_status(zhp->zpool_config, B_FALSE);
+
+	if (ret >= NMSGID)
+		*msgid = NULL;
+	else
+		*msgid = zfs_msgid_table[ret];
+
+	return (ret);
+}
+
+zpool_status_t
+zpool_import_status(nvlist_t *config, char **msgid)
+{
+	zpool_status_t ret = check_status(config, B_TRUE);
+
+	if (ret >= NMSGID)
+		*msgid = NULL;
+	else
+		*msgid = zfs_msgid_table[ret];
+
+	return (ret);
+}
+
+static void
+dump_ddt_stat(const ddt_stat_t *dds, int h)
+{
+	char refcnt[6];
+	char blocks[6], lsize[6], psize[6], dsize[6];
+	char ref_blocks[6], ref_lsize[6], ref_psize[6], ref_dsize[6];
+
+	if (dds == NULL || dds->dds_blocks == 0)
+		return;
+
+	if (h == -1)
+		(void) strcpy(refcnt, "Total");
+	else
+		zfs_nicenum(1ULL << h, refcnt, sizeof (refcnt));
+
+	zfs_nicenum(dds->dds_blocks, blocks, sizeof (blocks));
+	zfs_nicenum(dds->dds_lsize, lsize, sizeof (lsize));
+	zfs_nicenum(dds->dds_psize, psize, sizeof (psize));
+	zfs_nicenum(dds->dds_dsize, dsize, sizeof (dsize));
+	zfs_nicenum(dds->dds_ref_blocks, ref_blocks, sizeof (ref_blocks));
+	zfs_nicenum(dds->dds_ref_lsize, ref_lsize, sizeof (ref_lsize));
+	zfs_nicenum(dds->dds_ref_psize, ref_psize, sizeof (ref_psize));
+	zfs_nicenum(dds->dds_ref_dsize, ref_dsize, sizeof (ref_dsize));
+
+	(void) printf("%6s   %6s   %5s   %5s   %5s   %6s   %5s   %5s   %5s\n",
+	    refcnt,
+	    blocks, lsize, psize, dsize,
+	    ref_blocks, ref_lsize, ref_psize, ref_dsize);
+}
+
+/*
+ * Print the DDT histogram and the column totals.
+ */
+void
+zpool_dump_ddt(const ddt_stat_t *dds_total, const ddt_histogram_t *ddh)
+{
+	int h;
+
+	(void) printf("\n");
+
+	(void) printf("bucket   "
+	    "           allocated             "
+	    "          referenced          \n");
+	(void) printf("______   "
+	    "______________________________   "
+	    "______________________________\n");
+
+	(void) printf("%6s   %6s   %5s   %5s   %5s   %6s   %5s   %5s   %5s\n",
+	    "refcnt",
+	    "blocks", "LSIZE", "PSIZE", "DSIZE",
+	    "blocks", "LSIZE", "PSIZE", "DSIZE");
+
+	(void) printf("%6s   %6s   %5s   %5s   %5s   %6s   %5s   %5s   %5s\n",
+	    "------",
+	    "------", "-----", "-----", "-----",
+	    "------", "-----", "-----", "-----");
+
+	for (h = 0; h < 64; h++)
+		dump_ddt_stat(&ddh->ddh_stat[h], h);
+
+	dump_ddt_stat(dds_total, -1);
+
+	(void) printf("\n");
+}

lib/libzpool/common/kernel.c

@@ -0,0 +1,985 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2012 by Delphix. All rights reserved.
+ */
+
+#include <assert.h>
+#include <fcntl.h>
+#include <poll.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <zlib.h>
+#include <sys/spa.h>
+#include <sys/stat.h>
+#include <sys/processor.h>
+#include <sys/zfs_context.h>
+#include <sys/zmod.h>
+#include <sys/utsname.h>
+#include <sys/systeminfo.h>
+
+/*
+ * Emulation of kernel services in userland.
+ */
+
+int aok;
+uint64_t physmem;
+vnode_t *rootdir = (vnode_t *)0xabcd1234;
+char hw_serial[HW_HOSTID_LEN];
+vmem_t *zio_arena = NULL;
+
+struct utsname utsname = {
+	"userland", "libzpool", "1", "1", "na"
+};
+
+/* this only exists to have its address taken */
+struct proc p0;
+
+/*
+ * =========================================================================
+ * threads
+ * =========================================================================
+ */
+/*ARGSUSED*/
+kthread_t *
+zk_thread_create(void (*func)(), void *arg)
+{
+	thread_t tid;
+
+	VERIFY(thr_create(0, 0, (void *(*)(void *))func, arg, THR_DETACHED,
+	    &tid) == 0);
+
+	return ((void *)(uintptr_t)tid);
+}
+
+/*
+ * =========================================================================
+ * kstats
+ * =========================================================================
+ */
+/*ARGSUSED*/
+kstat_t *
+kstat_create(char *module, int instance, char *name, char *class,
+    uchar_t type, ulong_t ndata, uchar_t ks_flag)
+{
+	return (NULL);
+}
+
+/*ARGSUSED*/
+void
+kstat_install(kstat_t *ksp)
+{}
+
+/*ARGSUSED*/
+void
+kstat_delete(kstat_t *ksp)
+{}
+
+/*
+ * =========================================================================
+ * mutexes
+ * =========================================================================
+ */
+void
+zmutex_init(kmutex_t *mp)
+{
+	mp->m_owner = NULL;
+	mp->initialized = B_TRUE;
+	(void) _mutex_init(&mp->m_lock, USYNC_THREAD, NULL);
+}
+
+void
+zmutex_destroy(kmutex_t *mp)
+{
+	ASSERT(mp->initialized == B_TRUE);
+	ASSERT(mp->m_owner == NULL);
+	(void) _mutex_destroy(&(mp)->m_lock);
+	mp->m_owner = (void *)-1UL;
+	mp->initialized = B_FALSE;
+}
+
+void
+mutex_enter(kmutex_t *mp)
+{
+	ASSERT(mp->initialized == B_TRUE);
+	ASSERT(mp->m_owner != (void *)-1UL);
+	ASSERT(mp->m_owner != curthread);
+	VERIFY(mutex_lock(&mp->m_lock) == 0);
+	ASSERT(mp->m_owner == NULL);
+	mp->m_owner = curthread;
+}
+
+int
+mutex_tryenter(kmutex_t *mp)
+{
+	ASSERT(mp->initialized == B_TRUE);
+	ASSERT(mp->m_owner != (void *)-1UL);
+	if (0 == mutex_trylock(&mp->m_lock)) {
+		ASSERT(mp->m_owner == NULL);
+		mp->m_owner = curthread;
+		return (1);
+	} else {
+		return (0);
+	}
+}
+
+void
+mutex_exit(kmutex_t *mp)
+{
+	ASSERT(mp->initialized == B_TRUE);
+	ASSERT(mutex_owner(mp) == curthread);
+	mp->m_owner = NULL;
+	VERIFY(mutex_unlock(&mp->m_lock) == 0);
+}
+
+void *
+mutex_owner(kmutex_t *mp)
+{
+	ASSERT(mp->initialized == B_TRUE);
+	return (mp->m_owner);
+}
+
+/*
+ * =========================================================================
+ * rwlocks
+ * =========================================================================
+ */
+/*ARGSUSED*/
+void
+rw_init(krwlock_t *rwlp, char *name, int type, void *arg)
+{
+	rwlock_init(&rwlp->rw_lock, USYNC_THREAD, NULL);
+	rwlp->rw_owner = NULL;
+	rwlp->initialized = B_TRUE;
+}
+
+void
+rw_destroy(krwlock_t *rwlp)
+{
+	rwlock_destroy(&rwlp->rw_lock);
+	rwlp->rw_owner = (void *)-1UL;
+	rwlp->initialized = B_FALSE;
+}
+
+void
+rw_enter(krwlock_t *rwlp, krw_t rw)
+{
+	ASSERT(!RW_LOCK_HELD(rwlp));
+	ASSERT(rwlp->initialized == B_TRUE);
+	ASSERT(rwlp->rw_owner != (void *)-1UL);
+	ASSERT(rwlp->rw_owner != curthread);
+
+	if (rw == RW_READER)
+		VERIFY(rw_rdlock(&rwlp->rw_lock) == 0);
+	else
+		VERIFY(rw_wrlock(&rwlp->rw_lock) == 0);
+
+	rwlp->rw_owner = curthread;
+}
+
+void
+rw_exit(krwlock_t *rwlp)
+{
+	ASSERT(rwlp->initialized == B_TRUE);
+	ASSERT(rwlp->rw_owner != (void *)-1UL);
+
+	rwlp->rw_owner = NULL;
+	VERIFY(rw_unlock(&rwlp->rw_lock) == 0);
+}
+
+int
+rw_tryenter(krwlock_t *rwlp, krw_t rw)
+{
+	int rv;
+
+	ASSERT(rwlp->initialized == B_TRUE);
+	ASSERT(rwlp->rw_owner != (void *)-1UL);
+
+	if (rw == RW_READER)
+		rv = rw_tryrdlock(&rwlp->rw_lock);
+	else
+		rv = rw_trywrlock(&rwlp->rw_lock);
+
+	if (rv == 0) {
+		rwlp->rw_owner = curthread;
+		return (1);
+	}
+
+	return (0);
+}
+
+/*ARGSUSED*/
+int
+rw_tryupgrade(krwlock_t *rwlp)
+{
+	ASSERT(rwlp->initialized == B_TRUE);
+	ASSERT(rwlp->rw_owner != (void *)-1UL);
+
+	return (0);
+}
+
+/*
+ * =========================================================================
+ * condition variables
+ * =========================================================================
+ */
+/*ARGSUSED*/
+void
+cv_init(kcondvar_t *cv, char *name, int type, void *arg)
+{
+	VERIFY(cond_init(cv, type, NULL) == 0);
+}
+
+void
+cv_destroy(kcondvar_t *cv)
+{
+	VERIFY(cond_destroy(cv) == 0);
+}
+
+void
+cv_wait(kcondvar_t *cv, kmutex_t *mp)
+{
+	ASSERT(mutex_owner(mp) == curthread);
+	mp->m_owner = NULL;
+	int ret = cond_wait(cv, &mp->m_lock);
+	VERIFY(ret == 0 || ret == EINTR);
+	mp->m_owner = curthread;
+}
+
+clock_t
+cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime)
+{
+	int error;
+	timestruc_t ts;
+	clock_t delta;
+
+top:
+	delta = abstime - ddi_get_lbolt();
+	if (delta <= 0)
+		return (-1);
+
+	ts.tv_sec = delta / hz;
+	ts.tv_nsec = (delta % hz) * (NANOSEC / hz);
+
+	ASSERT(mutex_owner(mp) == curthread);
+	mp->m_owner = NULL;
+	error = cond_reltimedwait(cv, &mp->m_lock, &ts);
+	mp->m_owner = curthread;
+
+	if (error == ETIME)
+		return (-1);
+
+	if (error == EINTR)
+		goto top;
+
+	ASSERT(error == 0);
+
+	return (1);
+}
+
+void
+cv_signal(kcondvar_t *cv)
+{
+	VERIFY(cond_signal(cv) == 0);
+}
+
+void
+cv_broadcast(kcondvar_t *cv)
+{
+	VERIFY(cond_broadcast(cv) == 0);
+}
+
+/*
+ * =========================================================================
+ * vnode operations
+ * =========================================================================
+ */
+/*
+ * Note: for the xxxat() versions of these functions, we assume that the
+ * starting vp is always rootdir (which is true for spa_directory.c, the only
+ * ZFS consumer of these interfaces).  We assert this is true, and then emulate
+ * them by adding '/' in front of the path.
+ */
+
+/*ARGSUSED*/
+int
+vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3)
+{
+	int fd;
+	vnode_t *vp;
+	int old_umask;
+	char realpath[MAXPATHLEN];
+	struct stat64 st;
+
+	/*
+	 * If we're accessing a real disk from userland, we need to use
+	 * the character interface to avoid caching.  This is particularly
+	 * important if we're trying to look at a real in-kernel storage
+	 * pool from userland, e.g. via zdb, because otherwise we won't
+	 * see the changes occurring under the segmap cache.
+	 * On the other hand, the stupid character device returns zero
+	 * for its size.  So -- gag -- we open the block device to get
+	 * its size, and remember it for subsequent VOP_GETATTR().
+	 */
+	if (strncmp(path, "/dev/", 5) == 0) {
+		char *dsk;
+		fd = open64(path, O_RDONLY);
+		if (fd == -1)
+			return (errno);
+		if (fstat64(fd, &st) == -1) {
+			close(fd);
+			return (errno);
+		}
+		close(fd);
+		(void) sprintf(realpath, "%s", path);
+		dsk = strstr(path, "/dsk/");
+		if (dsk != NULL)
+			(void) sprintf(realpath + (dsk - path) + 1, "r%s",
+			    dsk + 1);
+	} else {
+		(void) sprintf(realpath, "%s", path);
+		if (!(flags & FCREAT) && stat64(realpath, &st) == -1)
+			return (errno);
+	}
+
+	if (flags & FCREAT)
+		old_umask = umask(0);
+
+	/*
+	 * The construct 'flags - FREAD' conveniently maps combinations of
+	 * FREAD and FWRITE to the corresponding O_RDONLY, O_WRONLY, and O_RDWR.
+	 */
+	fd = open64(realpath, flags - FREAD, mode);
+
+	if (flags & FCREAT)
+		(void) umask(old_umask);
+
+	if (fd == -1)
+		return (errno);
+
+	if (fstat64(fd, &st) == -1) {
+		close(fd);
+		return (errno);
+	}
+
+	(void) fcntl(fd, F_SETFD, FD_CLOEXEC);
+
+	*vpp = vp = umem_zalloc(sizeof (vnode_t), UMEM_NOFAIL);
+
+	vp->v_fd = fd;
+	vp->v_size = st.st_size;
+	vp->v_path = spa_strdup(path);
+
+	return (0);
+}
+
+/*ARGSUSED*/
+int
+vn_openat(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2,
+    int x3, vnode_t *startvp, int fd)
+{
+	char *realpath = umem_alloc(strlen(path) + 2, UMEM_NOFAIL);
+	int ret;
+
+	ASSERT(startvp == rootdir);
+	(void) sprintf(realpath, "/%s", path);
+
+	/* fd ignored for now, need if want to simulate nbmand support */
+	ret = vn_open(realpath, x1, flags, mode, vpp, x2, x3);
+
+	umem_free(realpath, strlen(path) + 2);
+
+	return (ret);
+}
+
+/*ARGSUSED*/
+int
+vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset,
+	int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp)
+{
+	ssize_t iolen, split;
+
+	if (uio == UIO_READ) {
+		iolen = pread64(vp->v_fd, addr, len, offset);
+	} else {
+		/*
+		 * To simulate partial disk writes, we split writes into two
+		 * system calls so that the process can be killed in between.
+		 */
+		int sectors = len >> SPA_MINBLOCKSHIFT;
+		split = (sectors > 0 ? rand() % sectors : 0) <<
+		    SPA_MINBLOCKSHIFT;
+		iolen = pwrite64(vp->v_fd, addr, split, offset);
+		iolen += pwrite64(vp->v_fd, (char *)addr + split,
+		    len - split, offset + split);
+	}
+
+	if (iolen == -1)
+		return (errno);
+	if (residp)
+		*residp = len - iolen;
+	else if (iolen != len)
+		return (EIO);
+	return (0);
+}
+
+void
+vn_close(vnode_t *vp)
+{
+	close(vp->v_fd);
+	spa_strfree(vp->v_path);
+	umem_free(vp, sizeof (vnode_t));
+}
+
+/*
+ * At a minimum we need to update the size since vdev_reopen()
+ * will no longer call vn_openat().
+ */
+int
+fop_getattr(vnode_t *vp, vattr_t *vap)
+{
+	struct stat64 st;
+
+	if (fstat64(vp->v_fd, &st) == -1) {
+		close(vp->v_fd);
+		return (errno);
+	}
+
+	vap->va_size = st.st_size;
+	return (0);
+}
+
+#ifdef ZFS_DEBUG
+
+/*
+ * =========================================================================
+ * Figure out which debugging statements to print
+ * =========================================================================
+ */
+
+static char *dprintf_string;
+static int dprintf_print_all;
+
+int
+dprintf_find_string(const char *string)
+{
+	char *tmp_str = dprintf_string;
+	int len = strlen(string);
+
+	/*
+	 * Find out if this is a string we want to print.
+	 * String format: file1.c,function_name1,file2.c,file3.c
+	 */
+
+	while (tmp_str != NULL) {
+		if (strncmp(tmp_str, string, len) == 0 &&
+		    (tmp_str[len] == ',' || tmp_str[len] == '\0'))
+			return (1);
+		tmp_str = strchr(tmp_str, ',');
+		if (tmp_str != NULL)
+			tmp_str++; /* Get rid of , */
+	}
+	return (0);
+}
+
+void
+dprintf_setup(int *argc, char **argv)
+{
+	int i, j;
+
+	/*
+	 * Debugging can be specified two ways: by setting the
+	 * environment variable ZFS_DEBUG, or by including a
+	 * "debug=..."  argument on the command line.  The command
+	 * line setting overrides the environment variable.
+	 */
+
+	for (i = 1; i < *argc; i++) {
+		int len = strlen("debug=");
+		/* First look for a command line argument */
+		if (strncmp("debug=", argv[i], len) == 0) {
+			dprintf_string = argv[i] + len;
+			/* Remove from args */
+			for (j = i; j < *argc; j++)
+				argv[j] = argv[j+1];
+			argv[j] = NULL;
+			(*argc)--;
+		}
+	}
+
+	if (dprintf_string == NULL) {
+		/* Look for ZFS_DEBUG environment variable */
+		dprintf_string = getenv("ZFS_DEBUG");
+	}
+
+	/*
+	 * Are we just turning on all debugging?
+	 */
+	if (dprintf_find_string("on"))
+		dprintf_print_all = 1;
+}
+
+/*
+ * =========================================================================
+ * debug printfs
+ * =========================================================================
+ */
+void
+__dprintf(const char *file, const char *func, int line, const char *fmt, ...)
+{
+	const char *newfile;
+	va_list adx;
+
+	/*
+	 * Get rid of annoying "../common/" prefix to filename.
+	 */
+	newfile = strrchr(file, '/');
+	if (newfile != NULL) {
+		newfile = newfile + 1; /* Get rid of leading / */
+	} else {
+		newfile = file;
+	}
+
+	if (dprintf_print_all ||
+	    dprintf_find_string(newfile) ||
+	    dprintf_find_string(func)) {
+		/* Print out just the function name if requested */
+		flockfile(stdout);
+		if (dprintf_find_string("pid"))
+			(void) printf("%d ", getpid());
+		if (dprintf_find_string("tid"))
+			(void) printf("%u ", thr_self());
+		if (dprintf_find_string("cpu"))
+			(void) printf("%u ", getcpuid());
+		if (dprintf_find_string("time"))
+			(void) printf("%llu ", gethrtime());
+		if (dprintf_find_string("long"))
+			(void) printf("%s, line %d: ", newfile, line);
+		(void) printf("%s: ", func);
+		va_start(adx, fmt);
+		(void) vprintf(fmt, adx);
+		va_end(adx);
+		funlockfile(stdout);
+	}
+}
+
+#endif /* ZFS_DEBUG */
+
+/*
+ * =========================================================================
+ * cmn_err() and panic()
+ * =========================================================================
+ */
+static char ce_prefix[CE_IGNORE][10] = { "", "NOTICE: ", "WARNING: ", "" };
+static char ce_suffix[CE_IGNORE][2] = { "", "\n", "\n", "" };
+
+void
+vpanic(const char *fmt, va_list adx)
+{
+	(void) fprintf(stderr, "error: ");
+	(void) vfprintf(stderr, fmt, adx);
+	(void) fprintf(stderr, "\n");
+
+	abort();	/* think of it as a "user-level crash dump" */
+}
+
+void
+panic(const char *fmt, ...)
+{
+	va_list adx;
+
+	va_start(adx, fmt);
+	vpanic(fmt, adx);
+	va_end(adx);
+}
+
+void
+vcmn_err(int ce, const char *fmt, va_list adx)
+{
+	if (ce == CE_PANIC)
+		vpanic(fmt, adx);
+	if (ce != CE_NOTE) {	/* suppress noise in userland stress testing */
+		(void) fprintf(stderr, "%s", ce_prefix[ce]);
+		(void) vfprintf(stderr, fmt, adx);
+		(void) fprintf(stderr, "%s", ce_suffix[ce]);
+	}
+}
+
+/*PRINTFLIKE2*/
+void
+cmn_err(int ce, const char *fmt, ...)
+{
+	va_list adx;
+
+	va_start(adx, fmt);
+	vcmn_err(ce, fmt, adx);
+	va_end(adx);
+}
+
+/*
+ * =========================================================================
+ * kobj interfaces
+ * =========================================================================
+ */
+struct _buf *
+kobj_open_file(char *name)
+{
+	struct _buf *file;
+	vnode_t *vp;
+
+	/* set vp as the _fd field of the file */
+	if (vn_openat(name, UIO_SYSSPACE, FREAD, 0, &vp, 0, 0, rootdir,
+	    -1) != 0)
+		return ((void *)-1UL);
+
+	file = umem_zalloc(sizeof (struct _buf), UMEM_NOFAIL);
+	file->_fd = (intptr_t)vp;
+	return (file);
+}
+
+int
+kobj_read_file(struct _buf *file, char *buf, unsigned size, unsigned off)
+{
+	ssize_t resid;
+
+	vn_rdwr(UIO_READ, (vnode_t *)file->_fd, buf, size, (offset_t)off,
+	    UIO_SYSSPACE, 0, 0, 0, &resid);
+
+	return (size - resid);
+}
+
+void
+kobj_close_file(struct _buf *file)
+{
+	vn_close((vnode_t *)file->_fd);
+	umem_free(file, sizeof (struct _buf));
+}
+
+int
+kobj_get_filesize(struct _buf *file, uint64_t *size)
+{
+	struct stat64 st;
+	vnode_t *vp = (vnode_t *)file->_fd;
+
+	if (fstat64(vp->v_fd, &st) == -1) {
+		vn_close(vp);
+		return (errno);
+	}
+	*size = st.st_size;
+	return (0);
+}
+
+/*
+ * =========================================================================
+ * misc routines
+ * =========================================================================
+ */
+
+void
+delay(clock_t ticks)
+{
+	poll(0, 0, ticks * (1000 / hz));
+}
+
+/*
+ * Find highest one bit set.
+ *	Returns bit number + 1 of highest bit that is set, otherwise returns 0.
+ * High order bit is 31 (or 63 in _LP64 kernel).
+ */
+int
+highbit(ulong_t i)
+{
+	register int h = 1;
+
+	if (i == 0)
+		return (0);
+#ifdef _LP64
+	if (i & 0xffffffff00000000ul) {
+		h += 32; i >>= 32;
+	}
+#endif
+	if (i & 0xffff0000) {
+		h += 16; i >>= 16;
+	}
+	if (i & 0xff00) {
+		h += 8; i >>= 8;
+	}
+	if (i & 0xf0) {
+		h += 4; i >>= 4;
+	}
+	if (i & 0xc) {
+		h += 2; i >>= 2;
+	}
+	if (i & 0x2) {
+		h += 1;
+	}
+	return (h);
+}
+
+static int random_fd = -1, urandom_fd = -1;
+
+static int
+random_get_bytes_common(uint8_t *ptr, size_t len, int fd)
+{
+	size_t resid = len;
+	ssize_t bytes;
+
+	ASSERT(fd != -1);
+
+	while (resid != 0) {
+		bytes = read(fd, ptr, resid);
+		ASSERT3S(bytes, >=, 0);
+		ptr += bytes;
+		resid -= bytes;
+	}
+
+	return (0);
+}
+
+int
+random_get_bytes(uint8_t *ptr, size_t len)
+{
+	return (random_get_bytes_common(ptr, len, random_fd));
+}
+
+int
+random_get_pseudo_bytes(uint8_t *ptr, size_t len)
+{
+	return (random_get_bytes_common(ptr, len, urandom_fd));
+}
+
+int
+ddi_strtoul(const char *hw_serial, char **nptr, int base, unsigned long *result)
+{
+	char *end;
+
+	*result = strtoul(hw_serial, &end, base);
+	if (*result == 0)
+		return (errno);
+	return (0);
+}
+
+int
+ddi_strtoull(const char *str, char **nptr, int base, u_longlong_t *result)
+{
+	char *end;
+
+	*result = strtoull(str, &end, base);
+	if (*result == 0)
+		return (errno);
+	return (0);
+}
+
+/*
+ * =========================================================================
+ * kernel emulation setup & teardown
+ * =========================================================================
+ */
+static int
+umem_out_of_memory(void)
+{
+	char errmsg[] = "out of memory -- generating core dump\n";
+
+	write(fileno(stderr), errmsg, sizeof (errmsg));
+	abort();
+	return (0);
+}
+
+void
+kernel_init(int mode)
+{
+	umem_nofail_callback(umem_out_of_memory);
+
+	physmem = sysconf(_SC_PHYS_PAGES);
+
+	dprintf("physmem = %llu pages (%.2f GB)\n", physmem,
+	    (double)physmem * sysconf(_SC_PAGE_SIZE) / (1ULL << 30));
+
+	(void) snprintf(hw_serial, sizeof (hw_serial), "%ld",
+	    (mode & FWRITE) ? gethostid() : 0);
+
+	VERIFY((random_fd = open("/dev/random", O_RDONLY)) != -1);
+	VERIFY((urandom_fd = open("/dev/urandom", O_RDONLY)) != -1);
+
+	system_taskq_init();
+
+	spa_init(mode);
+}
+
+void
+kernel_fini(void)
+{
+	spa_fini();
+
+	system_taskq_fini();
+
+	close(random_fd);
+	close(urandom_fd);
+
+	random_fd = -1;
+	urandom_fd = -1;
+}
+
+int
+z_uncompress(void *dst, size_t *dstlen, const void *src, size_t srclen)
+{
+	int ret;
+	uLongf len = *dstlen;
+
+	if ((ret = uncompress(dst, &len, src, srclen)) == Z_OK)
+		*dstlen = (size_t)len;
+
+	return (ret);
+}
+
+int
+z_compress_level(void *dst, size_t *dstlen, const void *src, size_t srclen,
+    int level)
+{
+	int ret;
+	uLongf len = *dstlen;
+
+	if ((ret = compress2(dst, &len, src, srclen, level)) == Z_OK)
+		*dstlen = (size_t)len;
+
+	return (ret);
+}
+
+uid_t
+crgetuid(cred_t *cr)
+{
+	return (0);
+}
+
+gid_t
+crgetgid(cred_t *cr)
+{
+	return (0);
+}
+
+int
+crgetngroups(cred_t *cr)
+{
+	return (0);
+}
+
+gid_t *
+crgetgroups(cred_t *cr)
+{
+	return (NULL);
+}
+
+int
+zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr)
+{
+	return (0);
+}
+
+int
+zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr)
+{
+	return (0);
+}
+
+int
+zfs_secpolicy_destroy_perms(const char *name, cred_t *cr)
+{
+	return (0);
+}
+
+ksiddomain_t *
+ksid_lookupdomain(const char *dom)
+{
+	ksiddomain_t *kd;
+
+	kd = umem_zalloc(sizeof (ksiddomain_t), UMEM_NOFAIL);
+	kd->kd_name = spa_strdup(dom);
+	return (kd);
+}
+
+void
+ksiddomain_rele(ksiddomain_t *ksid)
+{
+	spa_strfree(ksid->kd_name);
+	umem_free(ksid, sizeof (ksiddomain_t));
+}
+
+/*
+ * Do not change the length of the returned string; it must be freed
+ * with strfree().
+ */
+char *
+kmem_asprintf(const char *fmt, ...)
+{
+	int size;
+	va_list adx;
+	char *buf;
+
+	va_start(adx, fmt);
+	size = vsnprintf(NULL, 0, fmt, adx) + 1;
+	va_end(adx);
+
+	buf = kmem_alloc(size, KM_SLEEP);
+
+	va_start(adx, fmt);
+	size = vsnprintf(buf, size, fmt, adx);
+	va_end(adx);
+
+	return (buf);
+}
+
+/* ARGSUSED */
+int
+zfs_onexit_fd_hold(int fd, minor_t *minorp)
+{
+	*minorp = 0;
+	return (0);
+}
+
+/* ARGSUSED */
+void
+zfs_onexit_fd_rele(int fd)
+{
+}
+
+/* ARGSUSED */
+int
+zfs_onexit_add_cb(minor_t minor, void (*func)(void *), void *data,
+    uint64_t *action_handle)
+{
+	return (0);
+}
+
+/* ARGSUSED */
+int
+zfs_onexit_del_cb(minor_t minor, uint64_t action_handle, boolean_t fire)
+{
+	return (0);
+}
+
+/* ARGSUSED */
+int
+zfs_onexit_cb_data(minor_t minor, uint64_t action_handle, void **data)
+{
+	return (0);
+}

lib/libzpool/common/sys/zfs_context.h

@@ -0,0 +1,631 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
+ * Copyright (c) 2012 by Delphix. All rights reserved.
+ * Copyright (c) 2012, Joyent, Inc. All rights reserved.
+ */
+
+#ifndef _SYS_ZFS_CONTEXT_H
+#define	_SYS_ZFS_CONTEXT_H
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+#define	_SYS_MUTEX_H
+#define	_SYS_RWLOCK_H
+#define	_SYS_CONDVAR_H
+#define	_SYS_SYSTM_H
+#define	_SYS_DEBUG_H
+#define	_SYS_T_LOCK_H
+#define	_SYS_VNODE_H
+#define	_SYS_VFS_H
+#define	_SYS_SUNDDI_H
+#define	_SYS_CALLB_H
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stddef.h>
+#include <stdarg.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <errno.h>
+#include <string.h>
+#include <strings.h>
+#include <synch.h>
+#include <thread.h>
+#include <assert.h>
+#include <alloca.h>
+#include <umem.h>
+#include <limits.h>
+#include <atomic.h>
+#include <dirent.h>
+#include <time.h>
+#include <libsysevent.h>
+#include <sys/note.h>
+#include <sys/types.h>
+#include <sys/cred.h>
+#include <sys/sysmacros.h>
+#include <sys/bitmap.h>
+#include <sys/resource.h>
+#include <sys/byteorder.h>
+#include <sys/list.h>
+#include <sys/uio.h>
+#include <sys/zfs_debug.h>
+#include <sys/sdt.h>
+#include <sys/kstat.h>
+#include <sys/u8_textprep.h>
+#include <sys/sysevent/eventdefs.h>
+#include <sys/sysevent/dev.h>
+#include <sys/sunddi.h>
+
+/*
+ * Debugging
+ */
+
+/*
+ * Note that we are not using the debugging levels.
+ */
+
+#define	CE_CONT		0	/* continuation		*/
+#define	CE_NOTE		1	/* notice		*/
+#define	CE_WARN		2	/* warning		*/
+#define	CE_PANIC	3	/* panic		*/
+#define	CE_IGNORE	4	/* print nothing	*/
+
+/*
+ * ZFS debugging
+ */
+
+#ifdef ZFS_DEBUG
+extern void dprintf_setup(int *argc, char **argv);
+#endif /* ZFS_DEBUG */
+
+extern void cmn_err(int, const char *, ...);
+extern void vcmn_err(int, const char *, __va_list);
+extern void panic(const char *, ...);
+extern void vpanic(const char *, __va_list);
+
+#define	fm_panic	panic
+
+extern int aok;
+
+/* This definition is copied from assert.h. */
+#if defined(__STDC__)
+#if __STDC_VERSION__ - 0 >= 199901L
+#define	zverify(EX) (void)((EX) || (aok) || \
+	(__assert_c99(#EX, __FILE__, __LINE__, __func__), 0))
+#else
+#define	zverify(EX) (void)((EX) || (aok) || \
+	(__assert(#EX, __FILE__, __LINE__), 0))
+#endif /* __STDC_VERSION__ - 0 >= 199901L */
+#else
+#define	zverify(EX) (void)((EX) || (aok) || \
+	(_assert("EX", __FILE__, __LINE__), 0))
+#endif	/* __STDC__ */
+
+
+#define	VERIFY	zverify
+#define	ASSERT	zverify
+#undef	assert
+#define	assert	zverify
+
+extern void __assert(const char *, const char *, int);
+
+#ifdef lint
+#define	VERIFY3_IMPL(x, y, z, t)	if (x == z) ((void)0)
+#else
+/* BEGIN CSTYLED */
+#define	VERIFY3_IMPL(LEFT, OP, RIGHT, TYPE) do { \
+	const TYPE __left = (TYPE)(LEFT); \
+	const TYPE __right = (TYPE)(RIGHT); \
+	if (!(__left OP __right) && (!aok)) { \
+		char *__buf = alloca(256); \
+		(void) snprintf(__buf, 256, "%s %s %s (0x%llx %s 0x%llx)", \
+			#LEFT, #OP, #RIGHT, \
+			(u_longlong_t)__left, #OP, (u_longlong_t)__right); \
+		__assert(__buf, __FILE__, __LINE__); \
+	} \
+_NOTE(CONSTCOND) } while (0)
+/* END CSTYLED */
+#endif /* lint */
+
+#define	VERIFY3S(x, y, z)	VERIFY3_IMPL(x, y, z, int64_t)
+#define	VERIFY3U(x, y, z)	VERIFY3_IMPL(x, y, z, uint64_t)
+#define	VERIFY3P(x, y, z)	VERIFY3_IMPL(x, y, z, uintptr_t)
+
+#ifdef NDEBUG
+#define	ASSERT3S(x, y, z)	((void)0)
+#define	ASSERT3U(x, y, z)	((void)0)
+#define	ASSERT3P(x, y, z)	((void)0)
+#else
+#define	ASSERT3S(x, y, z)	VERIFY3S(x, y, z)
+#define	ASSERT3U(x, y, z)	VERIFY3U(x, y, z)
+#define	ASSERT3P(x, y, z)	VERIFY3P(x, y, z)
+#endif
+
+/*
+ * DTrace SDT probes have different signatures in userland than they do in
+ * kernel.  If they're being used in kernel code, re-define them out of
+ * existence for their counterparts in libzpool.
+ */
+
+#ifdef DTRACE_PROBE
+#undef	DTRACE_PROBE
+#define	DTRACE_PROBE(a)	((void)0)
+#endif	/* DTRACE_PROBE */
+
+#ifdef DTRACE_PROBE1
+#undef	DTRACE_PROBE1
+#define	DTRACE_PROBE1(a, b, c)	((void)0)
+#endif	/* DTRACE_PROBE1 */
+
+#ifdef DTRACE_PROBE2
+#undef	DTRACE_PROBE2
+#define	DTRACE_PROBE2(a, b, c, d, e)	((void)0)
+#endif	/* DTRACE_PROBE2 */
+
+#ifdef DTRACE_PROBE3
+#undef	DTRACE_PROBE3
+#define	DTRACE_PROBE3(a, b, c, d, e, f, g)	((void)0)
+#endif	/* DTRACE_PROBE3 */
+
+#ifdef DTRACE_PROBE4
+#undef	DTRACE_PROBE4
+#define	DTRACE_PROBE4(a, b, c, d, e, f, g, h, i)	((void)0)
+#endif	/* DTRACE_PROBE4 */
+
+/*
+ * Threads
+ */
+#define	curthread	((void *)(uintptr_t)thr_self())
+
+typedef struct kthread kthread_t;
+
+#define	thread_create(stk, stksize, func, arg, len, pp, state, pri)	\
+	zk_thread_create(func, arg)
+#define	thread_exit() thr_exit(NULL)
+#define	thread_join(t)	panic("libzpool cannot join threads")
+
+#define	newproc(f, a, cid, pri, ctp, pid)	(ENOSYS)
+
+/* in libzpool, p0 exists only to have its address taken */
+struct proc {
+	uintptr_t	this_is_never_used_dont_dereference_it;
+};
+
+extern struct proc p0;
+#define	curproc		(&p0)
+
+#define	PS_NONE		-1
+
+extern kthread_t *zk_thread_create(void (*func)(), void *arg);
+
+#define	issig(why)	(FALSE)
+#define	ISSIG(thr, why)	(FALSE)
+
+/*
+ * Mutexes
+ */
+typedef struct kmutex {
+	void		*m_owner;
+	boolean_t	initialized;
+	mutex_t		m_lock;
+} kmutex_t;
+
+#define	MUTEX_DEFAULT	USYNC_THREAD
+#undef	MUTEX_HELD
+#undef	MUTEX_NOT_HELD
+#define	MUTEX_HELD(m) _mutex_held(&(m)->m_lock)
+#define	MUTEX_NOT_HELD(m) (!MUTEX_HELD(m))
+
+/*
+ * Argh -- we have to get cheesy here because the kernel and userland
+ * have different signatures for the same routine.
+ */
+extern int _mutex_init(mutex_t *mp, int type, void *arg);
+extern int _mutex_destroy(mutex_t *mp);
+
+#define	mutex_init(mp, b, c, d)		zmutex_init((kmutex_t *)(mp))
+#define	mutex_destroy(mp)		zmutex_destroy((kmutex_t *)(mp))
+
+extern void zmutex_init(kmutex_t *mp);
+extern void zmutex_destroy(kmutex_t *mp);
+extern void mutex_enter(kmutex_t *mp);
+extern void mutex_exit(kmutex_t *mp);
+extern int mutex_tryenter(kmutex_t *mp);
+extern void *mutex_owner(kmutex_t *mp);
+
+/*
+ * RW locks
+ */
+typedef struct krwlock {
+	void		*rw_owner;
+	boolean_t	initialized;
+	rwlock_t	rw_lock;
+} krwlock_t;
+
+typedef int krw_t;
+
+#define	RW_READER	0
+#define	RW_WRITER	1
+#define	RW_DEFAULT	USYNC_THREAD
+
+#undef RW_READ_HELD
+#define	RW_READ_HELD(x)		_rw_read_held(&(x)->rw_lock)
+
+#undef RW_WRITE_HELD
+#define	RW_WRITE_HELD(x)	_rw_write_held(&(x)->rw_lock)
+
+extern void rw_init(krwlock_t *rwlp, char *name, int type, void *arg);
+extern void rw_destroy(krwlock_t *rwlp);
+extern void rw_enter(krwlock_t *rwlp, krw_t rw);
+extern int rw_tryenter(krwlock_t *rwlp, krw_t rw);
+extern int rw_tryupgrade(krwlock_t *rwlp);
+extern void rw_exit(krwlock_t *rwlp);
+#define	rw_downgrade(rwlp) do { } while (0)
+
+extern uid_t crgetuid(cred_t *cr);
+extern gid_t crgetgid(cred_t *cr);
+extern int crgetngroups(cred_t *cr);
+extern gid_t *crgetgroups(cred_t *cr);
+
+/*
+ * Condition variables
+ */
+typedef cond_t kcondvar_t;
+
+#define	CV_DEFAULT	USYNC_THREAD
+
+extern void cv_init(kcondvar_t *cv, char *name, int type, void *arg);
+extern void cv_destroy(kcondvar_t *cv);
+extern void cv_wait(kcondvar_t *cv, kmutex_t *mp);
+extern clock_t cv_timedwait(kcondvar_t *cv, kmutex_t *mp, clock_t abstime);
+extern void cv_signal(kcondvar_t *cv);
+extern void cv_broadcast(kcondvar_t *cv);
+
+/*
+ * kstat creation, installation and deletion
+ */
+extern kstat_t *kstat_create(char *, int,
+    char *, char *, uchar_t, ulong_t, uchar_t);
+extern void kstat_install(kstat_t *);
+extern void kstat_delete(kstat_t *);
+
+/*
+ * Kernel memory
+ */
+#define	KM_SLEEP		UMEM_NOFAIL
+#define	KM_PUSHPAGE		KM_SLEEP
+#define	KM_NOSLEEP		UMEM_DEFAULT
+#define	KMC_NODEBUG		UMC_NODEBUG
+#define	KMC_NOTOUCH		0	/* not needed for userland caches */
+#define	kmem_alloc(_s, _f)	umem_alloc(_s, _f)
+#define	kmem_zalloc(_s, _f)	umem_zalloc(_s, _f)
+#define	kmem_free(_b, _s)	umem_free(_b, _s)
+#define	kmem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i) \
+	umem_cache_create(_a, _b, _c, _d, _e, _f, _g, _h, _i)
+#define	kmem_cache_destroy(_c)	umem_cache_destroy(_c)
+#define	kmem_cache_alloc(_c, _f) umem_cache_alloc(_c, _f)
+#define	kmem_cache_free(_c, _b)	umem_cache_free(_c, _b)
+#define	kmem_debugging()	0
+#define	kmem_cache_reap_now(_c)		/* nothing */
+#define	kmem_cache_set_move(_c, _cb)	/* nothing */
+#define	vmem_qcache_reap(_v)		/* nothing */
+#define	POINTER_INVALIDATE(_pp)		/* nothing */
+#define	POINTER_IS_VALID(_p)	0
+
+extern vmem_t *zio_arena;
+
+typedef umem_cache_t kmem_cache_t;
+
+typedef enum kmem_cbrc {
+	KMEM_CBRC_YES,
+	KMEM_CBRC_NO,
+	KMEM_CBRC_LATER,
+	KMEM_CBRC_DONT_NEED,
+	KMEM_CBRC_DONT_KNOW
+} kmem_cbrc_t;
+
+/*
+ * Task queues
+ */
+typedef struct taskq taskq_t;
+typedef uintptr_t taskqid_t;
+typedef void (task_func_t)(void *);
+
+typedef struct taskq_ent {
+	struct taskq_ent	*tqent_next;
+	struct taskq_ent	*tqent_prev;
+	task_func_t		*tqent_func;
+	void			*tqent_arg;
+	uintptr_t		tqent_flags;
+} taskq_ent_t;
+
+#define	TQENT_FLAG_PREALLOC	0x1	/* taskq_dispatch_ent used */
+
+#define	TASKQ_PREPOPULATE	0x0001
+#define	TASKQ_CPR_SAFE		0x0002	/* Use CPR safe protocol */
+#define	TASKQ_DYNAMIC		0x0004	/* Use dynamic thread scheduling */
+#define	TASKQ_THREADS_CPU_PCT	0x0008	/* Scale # threads by # cpus */
+#define	TASKQ_DC_BATCH		0x0010	/* Mark threads as batch */
+
+#define	TQ_SLEEP	KM_SLEEP	/* Can block for memory */
+#define	TQ_NOSLEEP	KM_NOSLEEP	/* cannot block for memory; may fail */
+#define	TQ_NOQUEUE	0x02		/* Do not enqueue if can't dispatch */
+#define	TQ_FRONT	0x08		/* Queue in front */
+
+
+extern taskq_t *system_taskq;
+
+extern taskq_t	*taskq_create(const char *, int, pri_t, int, int, uint_t);
+#define	taskq_create_proc(a, b, c, d, e, p, f) \
+	    (taskq_create(a, b, c, d, e, f))
+#define	taskq_create_sysdc(a, b, d, e, p, dc, f) \
+	    (taskq_create(a, b, maxclsyspri, d, e, f))
+extern taskqid_t taskq_dispatch(taskq_t *, task_func_t, void *, uint_t);
+extern void	taskq_dispatch_ent(taskq_t *, task_func_t, void *, uint_t,
+    taskq_ent_t *);
+extern void	taskq_destroy(taskq_t *);
+extern void	taskq_wait(taskq_t *);
+extern int	taskq_member(taskq_t *, void *);
+extern void	system_taskq_init(void);
+extern void	system_taskq_fini(void);
+
+#define	XVA_MAPSIZE	3
+#define	XVA_MAGIC	0x78766174
+
+/*
+ * vnodes
+ */
+typedef struct vnode {
+	uint64_t	v_size;
+	int		v_fd;
+	char		*v_path;
+} vnode_t;
+
+#define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
+
+typedef struct xoptattr {
+	timestruc_t	xoa_createtime;	/* Create time of file */
+	uint8_t		xoa_archive;
+	uint8_t		xoa_system;
+	uint8_t		xoa_readonly;
+	uint8_t		xoa_hidden;
+	uint8_t		xoa_nounlink;
+	uint8_t		xoa_immutable;
+	uint8_t		xoa_appendonly;
+	uint8_t		xoa_nodump;
+	uint8_t		xoa_settable;
+	uint8_t		xoa_opaque;
+	uint8_t		xoa_av_quarantined;
+	uint8_t		xoa_av_modified;
+	uint8_t		xoa_av_scanstamp[AV_SCANSTAMP_SZ];
+	uint8_t		xoa_reparse;
+	uint8_t		xoa_offline;
+	uint8_t		xoa_sparse;
+} xoptattr_t;
+
+typedef struct vattr {
+	uint_t		va_mask;	/* bit-mask of attributes */
+	u_offset_t	va_size;	/* file size in bytes */
+} vattr_t;
+
+
+typedef struct xvattr {
+	vattr_t		xva_vattr;	/* Embedded vattr structure */
+	uint32_t	xva_magic;	/* Magic Number */
+	uint32_t	xva_mapsize;	/* Size of attr bitmap (32-bit words) */
+	uint32_t	*xva_rtnattrmapp;	/* Ptr to xva_rtnattrmap[] */
+	uint32_t	xva_reqattrmap[XVA_MAPSIZE];	/* Requested attrs */
+	uint32_t	xva_rtnattrmap[XVA_MAPSIZE];	/* Returned attrs */
+	xoptattr_t	xva_xoptattrs;	/* Optional attributes */
+} xvattr_t;
+
+typedef struct vsecattr {
+	uint_t		vsa_mask;	/* See below */
+	int		vsa_aclcnt;	/* ACL entry count */
+	void		*vsa_aclentp;	/* pointer to ACL entries */
+	int		vsa_dfaclcnt;	/* default ACL entry count */
+	void		*vsa_dfaclentp;	/* pointer to default ACL entries */
+	size_t		vsa_aclentsz;	/* ACE size in bytes of vsa_aclentp */
+} vsecattr_t;
+
+#define	AT_TYPE		0x00001
+#define	AT_MODE		0x00002
+#define	AT_UID		0x00004
+#define	AT_GID		0x00008
+#define	AT_FSID		0x00010
+#define	AT_NODEID	0x00020
+#define	AT_NLINK	0x00040
+#define	AT_SIZE		0x00080
+#define	AT_ATIME	0x00100
+#define	AT_MTIME	0x00200
+#define	AT_CTIME	0x00400
+#define	AT_RDEV		0x00800
+#define	AT_BLKSIZE	0x01000
+#define	AT_NBLOCKS	0x02000
+#define	AT_SEQ		0x08000
+#define	AT_XVATTR	0x10000
+
+#define	CRCREAT		0
+
+extern int fop_getattr(vnode_t *vp, vattr_t *vap);
+
+#define	VOP_CLOSE(vp, f, c, o, cr, ct)	0
+#define	VOP_PUTPAGE(vp, of, sz, fl, cr, ct)	0
+#define	VOP_GETATTR(vp, vap, fl, cr, ct)  fop_getattr((vp), (vap));
+
+#define	VOP_FSYNC(vp, f, cr, ct)	fsync((vp)->v_fd)
+
+#define	VN_RELE(vp)	vn_close(vp)
+
+extern int vn_open(char *path, int x1, int oflags, int mode, vnode_t **vpp,
+    int x2, int x3);
+extern int vn_openat(char *path, int x1, int oflags, int mode, vnode_t **vpp,
+    int x2, int x3, vnode_t *vp, int fd);
+extern int vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len,
+    offset_t offset, int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp);
+extern void vn_close(vnode_t *vp);
+
+#define	vn_remove(path, x1, x2)		remove(path)
+#define	vn_rename(from, to, seg)	rename((from), (to))
+#define	vn_is_readonly(vp)		B_FALSE
+
+extern vnode_t *rootdir;
+
+#include <sys/file.h>		/* for FREAD, FWRITE, etc */
+
+/*
+ * Random stuff
+ */
+#define	ddi_get_lbolt()		(gethrtime() >> 23)
+#define	ddi_get_lbolt64()	(gethrtime() >> 23)
+#define	hz	119	/* frequency when using gethrtime() >> 23 for lbolt */
+
+extern void delay(clock_t ticks);
+
+#define	gethrestime_sec() time(NULL)
+#define	gethrestime(t) \
+	do {\
+		(t)->tv_sec = gethrestime_sec();\
+		(t)->tv_nsec = 0;\
+	} while (0);
+
+#define	max_ncpus	64
+
+#define	minclsyspri	60
+#define	maxclsyspri	99
+
+#define	CPU_SEQID	(thr_self() & (max_ncpus - 1))
+
+#define	kcred		NULL
+#define	CRED()		NULL
+
+#define	ptob(x)		((x) * PAGESIZE)
+
+extern uint64_t physmem;
+
+extern int highbit(ulong_t i);
+extern int random_get_bytes(uint8_t *ptr, size_t len);
+extern int random_get_pseudo_bytes(uint8_t *ptr, size_t len);
+
+extern void kernel_init(int);
+extern void kernel_fini(void);
+
+struct spa;
+extern void nicenum(uint64_t num, char *buf);
+extern void show_pool_stats(struct spa *);
+
+typedef struct callb_cpr {
+	kmutex_t	*cc_lockp;
+} callb_cpr_t;
+
+#define	CALLB_CPR_INIT(cp, lockp, func, name)	{		\
+	(cp)->cc_lockp = lockp;					\
+}
+
+#define	CALLB_CPR_SAFE_BEGIN(cp) {				\
+	ASSERT(MUTEX_HELD((cp)->cc_lockp));			\
+}
+
+#define	CALLB_CPR_SAFE_END(cp, lockp) {				\
+	ASSERT(MUTEX_HELD((cp)->cc_lockp));			\
+}
+
+#define	CALLB_CPR_EXIT(cp) {					\
+	ASSERT(MUTEX_HELD((cp)->cc_lockp));			\
+	mutex_exit((cp)->cc_lockp);				\
+}
+
+#define	zone_dataset_visible(x, y)	(1)
+#define	INGLOBALZONE(z)			(1)
+
+extern char *kmem_asprintf(const char *fmt, ...);
+#define	strfree(str) kmem_free((str), strlen(str)+1)
+
+/*
+ * Hostname information
+ */
+extern char hw_serial[];	/* for userland-emulated hostid access */
+extern int ddi_strtoul(const char *str, char **nptr, int base,
+    unsigned long *result);
+
+extern int ddi_strtoull(const char *str, char **nptr, int base,
+    u_longlong_t *result);
+
+/* ZFS Boot Related stuff. */
+
+struct _buf {
+	intptr_t	_fd;
+};
+
+struct bootstat {
+	uint64_t st_size;
+};
+
+typedef struct ace_object {
+	uid_t		a_who;
+	uint32_t	a_access_mask;
+	uint16_t	a_flags;
+	uint16_t	a_type;
+	uint8_t		a_obj_type[16];
+	uint8_t		a_inherit_obj_type[16];
+} ace_object_t;
+
+
+#define	ACE_ACCESS_ALLOWED_OBJECT_ACE_TYPE	0x05
+#define	ACE_ACCESS_DENIED_OBJECT_ACE_TYPE	0x06
+#define	ACE_SYSTEM_AUDIT_OBJECT_ACE_TYPE	0x07
+#define	ACE_SYSTEM_ALARM_OBJECT_ACE_TYPE	0x08
+
+extern struct _buf *kobj_open_file(char *name);
+extern int kobj_read_file(struct _buf *file, char *buf, unsigned size,
+    unsigned off);
+extern void kobj_close_file(struct _buf *file);
+extern int kobj_get_filesize(struct _buf *file, uint64_t *size);
+extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr);
+extern int zfs_secpolicy_rename_perms(const char *from, const char *to,
+    cred_t *cr);
+extern int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr);
+extern zoneid_t getzoneid(void);
+
+/* SID stuff */
+typedef struct ksiddomain {
+	uint_t	kd_ref;
+	uint_t	kd_len;
+	char	*kd_name;
+} ksiddomain_t;
+
+ksiddomain_t *ksid_lookupdomain(const char *);
+void ksiddomain_rele(ksiddomain_t *);
+
+#define	DDI_SLEEP	KM_SLEEP
+#define	ddi_log_sysevent(_a, _b, _c, _d, _e, _f, _g) \
+	sysevent_post_event(_c, _d, _b, "libzpool", _e, _f)
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _SYS_ZFS_CONTEXT_H */

lib/libzpool/common/taskq.c

@@ -0,0 +1,338 @@
+/*
+ * 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 2010 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+/*
+ * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
+ * Copyright 2012 Garrett D'Amore <garrett@damore.org>.  All rights reserved.
+ */
+
+#include <sys/zfs_context.h>
+
+int taskq_now;
+taskq_t *system_taskq;
+
+#define	TASKQ_ACTIVE	0x00010000
+
+struct taskq {
+	kmutex_t	tq_lock;
+	krwlock_t	tq_threadlock;
+	kcondvar_t	tq_dispatch_cv;
+	kcondvar_t	tq_wait_cv;
+	thread_t	*tq_threadlist;
+	int		tq_flags;
+	int		tq_active;
+	int		tq_nthreads;
+	int		tq_nalloc;
+	int		tq_minalloc;
+	int		tq_maxalloc;
+	kcondvar_t	tq_maxalloc_cv;
+	int		tq_maxalloc_wait;
+	taskq_ent_t	*tq_freelist;
+	taskq_ent_t	tq_task;
+};
+
+static taskq_ent_t *
+task_alloc(taskq_t *tq, int tqflags)
+{
+	taskq_ent_t *t;
+	int rv;
+
+again:	if ((t = tq->tq_freelist) != NULL && tq->tq_nalloc >= tq->tq_minalloc) {
+		tq->tq_freelist = t->tqent_next;
+	} else {
+		if (tq->tq_nalloc >= tq->tq_maxalloc) {
+			if (!(tqflags & KM_SLEEP))
+				return (NULL);
+
+			/*
+			 * We don't want to exceed tq_maxalloc, but we can't
+			 * wait for other tasks to complete (and thus free up
+			 * task structures) without risking deadlock with
+			 * the caller.  So, we just delay for one second
+			 * to throttle the allocation rate. If we have tasks
+			 * complete before one second timeout expires then
+			 * taskq_ent_free will signal us and we will
+			 * immediately retry the allocation.
+			 */
+			tq->tq_maxalloc_wait++;
+			rv = cv_timedwait(&tq->tq_maxalloc_cv,
+			    &tq->tq_lock, ddi_get_lbolt() + hz);
+			tq->tq_maxalloc_wait--;
+			if (rv > 0)
+				goto again;		/* signaled */
+		}
+		mutex_exit(&tq->tq_lock);
+
+		t = kmem_alloc(sizeof (taskq_ent_t), tqflags);
+
+		mutex_enter(&tq->tq_lock);
+		if (t != NULL)
+			tq->tq_nalloc++;
+	}
+	return (t);
+}
+
+static void
+task_free(taskq_t *tq, taskq_ent_t *t)
+{
+	if (tq->tq_nalloc <= tq->tq_minalloc) {
+		t->tqent_next = tq->tq_freelist;
+		tq->tq_freelist = t;
+	} else {
+		tq->tq_nalloc--;
+		mutex_exit(&tq->tq_lock);
+		kmem_free(t, sizeof (taskq_ent_t));
+		mutex_enter(&tq->tq_lock);
+	}
+
+	if (tq->tq_maxalloc_wait)
+		cv_signal(&tq->tq_maxalloc_cv);
+}
+
+taskqid_t
+taskq_dispatch(taskq_t *tq, task_func_t func, void *arg, uint_t tqflags)
+{
+	taskq_ent_t *t;
+
+	if (taskq_now) {
+		func(arg);
+		return (1);
+	}
+
+	mutex_enter(&tq->tq_lock);
+	ASSERT(tq->tq_flags & TASKQ_ACTIVE);
+	if ((t = task_alloc(tq, tqflags)) == NULL) {
+		mutex_exit(&tq->tq_lock);
+		return (0);
+	}
+	if (tqflags & TQ_FRONT) {
+		t->tqent_next = tq->tq_task.tqent_next;
+		t->tqent_prev = &tq->tq_task;
+	} else {
+		t->tqent_next = &tq->tq_task;
+		t->tqent_prev = tq->tq_task.tqent_prev;
+	}
+	t->tqent_next->tqent_prev = t;
+	t->tqent_prev->tqent_next = t;
+	t->tqent_func = func;
+	t->tqent_arg = arg;
+	t->tqent_flags = 0;
+	cv_signal(&tq->tq_dispatch_cv);
+	mutex_exit(&tq->tq_lock);
+	return (1);
+}
+
+void
+taskq_dispatch_ent(taskq_t *tq, task_func_t func, void *arg, uint_t flags,
+    taskq_ent_t *t)
+{
+	ASSERT(func != NULL);
+	ASSERT(!(tq->tq_flags & TASKQ_DYNAMIC));
+
+	/*
+	 * Mark it as a prealloc'd task.  This is important
+	 * to ensure that we don't free it later.
+	 */
+	t->tqent_flags |= TQENT_FLAG_PREALLOC;
+	/*
+	 * Enqueue the task to the underlying queue.
+	 */
+	mutex_enter(&tq->tq_lock);
+
+	if (flags & TQ_FRONT) {
+		t->tqent_next = tq->tq_task.tqent_next;
+		t->tqent_prev = &tq->tq_task;
+	} else {
+		t->tqent_next = &tq->tq_task;
+		t->tqent_prev = tq->tq_task.tqent_prev;
+	}
+	t->tqent_next->tqent_prev = t;
+	t->tqent_prev->tqent_next = t;
+	t->tqent_func = func;
+	t->tqent_arg = arg;
+	cv_signal(&tq->tq_dispatch_cv);
+	mutex_exit(&tq->tq_lock);
+}
+
+void
+taskq_wait(taskq_t *tq)
+{
+	mutex_enter(&tq->tq_lock);
+	while (tq->tq_task.tqent_next != &tq->tq_task || tq->tq_active != 0)
+		cv_wait(&tq->tq_wait_cv, &tq->tq_lock);
+	mutex_exit(&tq->tq_lock);
+}
+
+static void *
+taskq_thread(void *arg)
+{
+	taskq_t *tq = arg;
+	taskq_ent_t *t;
+	boolean_t prealloc;
+
+	mutex_enter(&tq->tq_lock);
+	while (tq->tq_flags & TASKQ_ACTIVE) {
+		if ((t = tq->tq_task.tqent_next) == &tq->tq_task) {
+			if (--tq->tq_active == 0)
+				cv_broadcast(&tq->tq_wait_cv);
+			cv_wait(&tq->tq_dispatch_cv, &tq->tq_lock);
+			tq->tq_active++;
+			continue;
+		}
+		t->tqent_prev->tqent_next = t->tqent_next;
+		t->tqent_next->tqent_prev = t->tqent_prev;
+		t->tqent_next = NULL;
+		t->tqent_prev = NULL;
+		prealloc = t->tqent_flags & TQENT_FLAG_PREALLOC;
+		mutex_exit(&tq->tq_lock);
+
+		rw_enter(&tq->tq_threadlock, RW_READER);
+		t->tqent_func(t->tqent_arg);
+		rw_exit(&tq->tq_threadlock);
+
+		mutex_enter(&tq->tq_lock);
+		if (!prealloc)
+			task_free(tq, t);
+	}
+	tq->tq_nthreads--;
+	cv_broadcast(&tq->tq_wait_cv);
+	mutex_exit(&tq->tq_lock);
+	return (NULL);
+}
+
+/*ARGSUSED*/
+taskq_t *
+taskq_create(const char *name, int nthreads, pri_t pri,
+	int minalloc, int maxalloc, uint_t flags)
+{
+	taskq_t *tq = kmem_zalloc(sizeof (taskq_t), KM_SLEEP);
+	int t;
+
+	if (flags & TASKQ_THREADS_CPU_PCT) {
+		int pct;
+		ASSERT3S(nthreads, >=, 0);
+		ASSERT3S(nthreads, <=, 100);
+		pct = MIN(nthreads, 100);
+		pct = MAX(pct, 0);
+
+		nthreads = (sysconf(_SC_NPROCESSORS_ONLN) * pct) / 100;
+		nthreads = MAX(nthreads, 1);	/* need at least 1 thread */
+	} else {
+		ASSERT3S(nthreads, >=, 1);
+	}
+
+	rw_init(&tq->tq_threadlock, NULL, RW_DEFAULT, NULL);
+	mutex_init(&tq->tq_lock, NULL, MUTEX_DEFAULT, NULL);
+	cv_init(&tq->tq_dispatch_cv, NULL, CV_DEFAULT, NULL);
+	cv_init(&tq->tq_wait_cv, NULL, CV_DEFAULT, NULL);
+	cv_init(&tq->tq_maxalloc_cv, NULL, CV_DEFAULT, NULL);
+	tq->tq_flags = flags | TASKQ_ACTIVE;
+	tq->tq_active = nthreads;
+	tq->tq_nthreads = nthreads;
+	tq->tq_minalloc = minalloc;
+	tq->tq_maxalloc = maxalloc;
+	tq->tq_task.tqent_next = &tq->tq_task;
+	tq->tq_task.tqent_prev = &tq->tq_task;
+	tq->tq_threadlist = kmem_alloc(nthreads * sizeof (thread_t), KM_SLEEP);
+
+	if (flags & TASKQ_PREPOPULATE) {
+		mutex_enter(&tq->tq_lock);
+		while (minalloc-- > 0)
+			task_free(tq, task_alloc(tq, KM_SLEEP));
+		mutex_exit(&tq->tq_lock);
+	}
+
+	for (t = 0; t < nthreads; t++)
+		(void) thr_create(0, 0, taskq_thread,
+		    tq, THR_BOUND, &tq->tq_threadlist[t]);
+
+	return (tq);
+}
+
+void
+taskq_destroy(taskq_t *tq)
+{
+	int t;
+	int nthreads = tq->tq_nthreads;
+
+	taskq_wait(tq);
+
+	mutex_enter(&tq->tq_lock);
+
+	tq->tq_flags &= ~TASKQ_ACTIVE;
+	cv_broadcast(&tq->tq_dispatch_cv);
+
+	while (tq->tq_nthreads != 0)
+		cv_wait(&tq->tq_wait_cv, &tq->tq_lock);
+
+	tq->tq_minalloc = 0;
+	while (tq->tq_nalloc != 0) {
+		ASSERT(tq->tq_freelist != NULL);
+		task_free(tq, task_alloc(tq, KM_SLEEP));
+	}
+
+	mutex_exit(&tq->tq_lock);
+
+	for (t = 0; t < nthreads; t++)
+		(void) thr_join(tq->tq_threadlist[t], NULL, NULL);
+
+	kmem_free(tq->tq_threadlist, nthreads * sizeof (thread_t));
+
+	rw_destroy(&tq->tq_threadlock);
+	mutex_destroy(&tq->tq_lock);
+	cv_destroy(&tq->tq_dispatch_cv);
+	cv_destroy(&tq->tq_wait_cv);
+	cv_destroy(&tq->tq_maxalloc_cv);
+
+	kmem_free(tq, sizeof (taskq_t));
+}
+
+int
+taskq_member(taskq_t *tq, void *t)
+{
+	int i;
+
+	if (taskq_now)
+		return (1);
+
+	for (i = 0; i < tq->tq_nthreads; i++)
+		if (tq->tq_threadlist[i] == (thread_t)(uintptr_t)t)
+			return (1);
+
+	return (0);
+}
+
+void
+system_taskq_init(void)
+{
+	system_taskq = taskq_create("system_taskq", 64, minclsyspri, 4, 512,
+	    TASKQ_DYNAMIC | TASKQ_PREPOPULATE);
+}
+
+void
+system_taskq_fini(void)
+{
+	taskq_destroy(system_taskq);
+	system_taskq = NULL; /* defensive */
+}

lib/libzpool/common/util.c

@@ -0,0 +1,155 @@
+/*
+ * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#include <assert.h>
+#include <sys/zfs_context.h>
+#include <sys/avl.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/spa.h>
+#include <sys/fs/zfs.h>
+#include <sys/refcount.h>
+
+/*
+ * Routines needed by more than one client of libzpool.
+ */
+
+void
+nicenum(uint64_t num, char *buf)
+{
+	uint64_t n = num;
+	int index = 0;
+	char u;
+
+	while (n >= 1024) {
+		n = (n + (1024 / 2)) / 1024; /* Round up or down */
+		index++;
+	}
+
+	u = " KMGTPE"[index];
+
+	if (index == 0) {
+		(void) sprintf(buf, "%llu", (u_longlong_t)n);
+	} else if (n < 10 && (num & (num - 1)) != 0) {
+		(void) sprintf(buf, "%.2f%c",
+		    (double)num / (1ULL << 10 * index), u);
+	} else if (n < 100 && (num & (num - 1)) != 0) {
+		(void) sprintf(buf, "%.1f%c",
+		    (double)num / (1ULL << 10 * index), u);
+	} else {
+		(void) sprintf(buf, "%llu%c", (u_longlong_t)n, u);
+	}
+}
+
+static void
+show_vdev_stats(const char *desc, const char *ctype, nvlist_t *nv, int indent)
+{
+	vdev_stat_t *vs;
+	vdev_stat_t v0 = { 0 };
+	uint64_t sec;
+	uint64_t is_log = 0;
+	nvlist_t **child;
+	uint_t c, children;
+	char used[6], avail[6];
+	char rops[6], wops[6], rbytes[6], wbytes[6], rerr[6], werr[6], cerr[6];
+	char *prefix = "";
+
+	if (indent == 0 && desc != NULL) {
+		(void) printf("                           "
+		    " capacity   operations   bandwidth  ---- errors ----\n");
+		(void) printf("description                "
+		    "used avail  read write  read write  read write cksum\n");
+	}
+
+	if (desc != NULL) {
+		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log);
+
+		if (is_log)
+			prefix = "log ";
+
+		if (nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
+		    (uint64_t **)&vs, &c) != 0)
+			vs = &v0;
+
+		sec = MAX(1, vs->vs_timestamp / NANOSEC);
+
+		nicenum(vs->vs_alloc, used);
+		nicenum(vs->vs_space - vs->vs_alloc, avail);
+		nicenum(vs->vs_ops[ZIO_TYPE_READ] / sec, rops);
+		nicenum(vs->vs_ops[ZIO_TYPE_WRITE] / sec, wops);
+		nicenum(vs->vs_bytes[ZIO_TYPE_READ] / sec, rbytes);
+		nicenum(vs->vs_bytes[ZIO_TYPE_WRITE] / sec, wbytes);
+		nicenum(vs->vs_read_errors, rerr);
+		nicenum(vs->vs_write_errors, werr);
+		nicenum(vs->vs_checksum_errors, cerr);
+
+		(void) printf("%*s%s%*s%*s%*s %5s %5s %5s %5s %5s %5s %5s\n",
+		    indent, "",
+		    prefix,
+		    indent + strlen(prefix) - 25 - (vs->vs_space ? 0 : 12),
+		    desc,
+		    vs->vs_space ? 6 : 0, vs->vs_space ? used : "",
+		    vs->vs_space ? 6 : 0, vs->vs_space ? avail : "",
+		    rops, wops, rbytes, wbytes, rerr, werr, cerr);
+	}
+
+	if (nvlist_lookup_nvlist_array(nv, ctype, &child, &children) != 0)
+		return;
+
+	for (c = 0; c < children; c++) {
+		nvlist_t *cnv = child[c];
+		char *cname, *tname;
+		uint64_t np;
+		if (nvlist_lookup_string(cnv, ZPOOL_CONFIG_PATH, &cname) &&
+		    nvlist_lookup_string(cnv, ZPOOL_CONFIG_TYPE, &cname))
+			cname = "<unknown>";
+		tname = calloc(1, strlen(cname) + 2);
+		(void) strcpy(tname, cname);
+		if (nvlist_lookup_uint64(cnv, ZPOOL_CONFIG_NPARITY, &np) == 0)
+			tname[strlen(tname)] = '0' + np;
+		show_vdev_stats(tname, ctype, cnv, indent + 2);
+		free(tname);
+	}
+}
+
+void
+show_pool_stats(spa_t *spa)
+{
+	nvlist_t *config, *nvroot;
+	char *name;
+
+	VERIFY(spa_get_stats(spa_name(spa), &config, NULL, 0) == 0);
+
+	VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
+	    &nvroot) == 0);
+	VERIFY(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
+	    &name) == 0);
+
+	show_vdev_stats(name, ZPOOL_CONFIG_CHILDREN, nvroot, 0);
+	show_vdev_stats(NULL, ZPOOL_CONFIG_L2CACHE, nvroot, 0);
+	show_vdev_stats(NULL, ZPOOL_CONFIG_SPARES, nvroot, 0);
+
+	nvlist_free(config);
+}

lib/pyzfs/common/__init__.py

@@ -0,0 +1,27 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""
+package which provides an administrative interface to ZFS
+"""

lib/pyzfs/common/allow.py

@@ -0,0 +1,396 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""This module implements the "zfs allow" and "zfs unallow" subcommands.
+The only public interface is the zfs.allow.do_allow() function."""
+
+import zfs.util
+import zfs.dataset
+import optparse
+import sys
+import pwd
+import grp
+import errno
+
+_ = zfs.util._
+
+class FSPerms(object):
+	"""This class represents all the permissions that are set on a
+	particular filesystem (not including those inherited)."""
+
+	__slots__ = "create", "sets", "local", "descend", "ld"
+	__repr__ = zfs.util.default_repr
+
+	def __init__(self, raw):
+		"""Create a FSPerms based on the dict of raw permissions
+		from zfs.ioctl.get_fsacl()."""
+		# set of perms
+		self.create = set()
+
+		# below are { "Ntype name": set(perms) }
+		# where N is a number that we just use for sorting,
+		# type is "user", "group", "everyone", or "" (for sets)
+		# name is a user, group, or set name, or "" (for everyone)
+		self.sets = dict()
+		self.local = dict()
+		self.descend = dict()
+		self.ld = dict()
+
+		# see the comment in dsl_deleg.c for the definition of whokey
+		for whokey in raw.keys():
+			perms = raw[whokey].keys()
+			whotypechr = whokey[0].lower()
+			ws = whokey[3:]
+			if whotypechr == "c":
+				self.create.update(perms)
+			elif whotypechr == "s":
+				nwho = "1" + ws
+				self.sets.setdefault(nwho, set()).update(perms)
+			else:
+				if whotypechr == "u":
+					try:
+						name = pwd.getpwuid(int(ws)).pw_name
+					except KeyError:
+						name = ws
+					nwho = "1user " + name
+				elif whotypechr == "g":
+					try:
+						name = grp.getgrgid(int(ws)).gr_name
+					except KeyError:
+						name = ws
+					nwho = "2group " + name
+				elif whotypechr == "e":
+					nwho = "3everyone"
+				else:
+					raise ValueError(whotypechr)
+
+				if whokey[1] == "l":
+					d = self.local
+				elif whokey[1] == "d":
+					d = self.descend
+				else:
+					raise ValueError(whokey[1])
+
+				d.setdefault(nwho, set()).update(perms)
+
+		# Find perms that are in both local and descend, and
+		# move them to ld.
+		for nwho in self.local:
+			if nwho not in self.descend:
+				continue
+			# note: these are set operations
+			self.ld[nwho] = self.local[nwho] & self.descend[nwho]
+			self.local[nwho] -= self.ld[nwho]
+			self.descend[nwho] -= self.ld[nwho]
+
+	@staticmethod
+	def __ldstr(d, header):
+		s = ""
+		for (nwho, perms) in sorted(d.items()):
+			# local and descend may have entries where perms
+			# is an empty set, due to consolidating all
+			# permissions into ld
+			if perms:
+				s += "\t%s %s\n" % \
+				    (nwho[1:], ",".join(sorted(perms)))
+		if s:
+			s = header + s
+		return s
+
+	def __str__(self):
+		s = self.__ldstr(self.sets, _("Permission sets:\n"))
+
+		if self.create:
+			s += _("Create time permissions:\n")
+			s += "\t%s\n" % ",".join(sorted(self.create))
+
+		s += self.__ldstr(self.local, _("Local permissions:\n"))
+		s += self.__ldstr(self.descend, _("Descendent permissions:\n"))
+		s += self.__ldstr(self.ld, _("Local+Descendent permissions:\n"))
+		return s.rstrip()
+
+def args_to_perms(parser, options, who, perms):
+	"""Return a dict of raw perms {"whostr" -> {"perm" -> None}}
+	based on the command-line input."""
+
+	# perms is not set if we are doing a "zfs unallow <who> <fs>" to
+	# remove all of someone's permissions
+	if perms:
+		setperms = dict(((p, None) for p in perms if p[0] == "@"))
+		baseperms = dict(((canonicalized_perm(p), None)
+		    for p in perms if p[0] != "@"))
+	else:
+		setperms = None
+		baseperms = None
+
+	d = dict()
+	
+	def storeperm(typechr, inheritchr, arg):
+		assert typechr in "ugecs"
+		assert inheritchr in "ld-"
+
+		def mkwhokey(t):
+			return "%c%c$%s" % (t, inheritchr, arg)
+
+		if baseperms or not perms:
+			d[mkwhokey(typechr)] = baseperms
+		if setperms or not perms:
+			d[mkwhokey(typechr.upper())] = setperms
+
+	def decodeid(w, toidfunc, fmt):
+		try:
+			return int(w)
+		except ValueError:
+			try:
+				return toidfunc(w)[2]
+			except KeyError:
+				parser.error(fmt % w)
+
+	if options.set:
+		storeperm("s", "-", who)
+	elif options.create:
+		storeperm("c", "-", "")
+	else:
+		for w in who:
+			if options.user:
+				id = decodeid(w, pwd.getpwnam,
+				    _("invalid user %s"))
+				typechr = "u"
+			elif options.group:
+				id = decodeid(w, grp.getgrnam,
+				    _("invalid group %s"))
+				typechr = "g"
+			elif w == "everyone":
+				id = ""
+				typechr = "e"
+			else:
+				try:
+					id = pwd.getpwnam(w)[2]
+					typechr = "u"
+				except KeyError:
+					try:
+						id = grp.getgrnam(w)[2]
+						typechr = "g"
+					except KeyError:
+						parser.error(_("invalid user/group %s") % w)
+			if options.local:
+				storeperm(typechr, "l", id)
+			if options.descend:
+				storeperm(typechr, "d", id)
+	return d
+
+perms_subcmd = dict(
+    create=_("Must also have the 'mount' ability"),
+    destroy=_("Must also have the 'mount' ability"),
+    snapshot="",
+    rollback="",
+    clone=_("""Must also have the 'create' ability and 'mount'
+\t\t\t\tability in the origin file system"""),
+    promote=_("""Must also have the 'mount'
+\t\t\t\tand 'promote' ability in the origin file system"""),
+    rename=_("""Must also have the 'mount' and 'create'
+\t\t\t\tability in the new parent"""),
+    receive=_("Must also have the 'mount' and 'create' ability"),
+    allow=_("Must also have the permission that is being\n\t\t\t\tallowed"),
+    mount=_("Allows mount/umount of ZFS datasets"),
+    share=_("Allows sharing file systems over NFS or SMB\n\t\t\t\tprotocols"),
+    send="",
+    hold=_("Allows adding a user hold to a snapshot"),
+    release=_("Allows releasing a user hold which\n\t\t\t\tmight destroy the snapshot"),
+    diff=_("Allows lookup of paths within a dataset,\n\t\t\t\tgiven an object number. Ordinary users need this\n\t\t\t\tin order to use zfs diff"),
+)
+
+perms_other = dict(
+    userprop=_("Allows changing any user property"),
+    userquota=_("Allows accessing any userquota@... property"),
+    groupquota=_("Allows accessing any groupquota@... property"),
+    userused=_("Allows reading any userused@... property"),
+    groupused=_("Allows reading any groupused@... property"),
+)
+
+def hasset(ds, setname):
+	"""Return True if the given setname (string) is defined for this
+	ds (Dataset)."""
+	# It would be nice to cache the result of get_fsacl().
+	for raw in ds.get_fsacl().values():
+		for whokey in raw.keys():
+			if whokey[0].lower() == "s" and whokey[3:] == setname:
+				return True
+	return False
+
+def canonicalized_perm(permname):
+	"""Return the canonical name (string) for this permission (string).
+	Raises ZFSError if it is not a valid permission."""
+	if permname in perms_subcmd.keys() or permname in perms_other.keys():
+		return permname
+	try:
+		return zfs.dataset.getpropobj(permname).name
+	except KeyError:
+		raise zfs.util.ZFSError(errno.EINVAL, permname,
+		    _("invalid permission"))
+		
+def print_perms():
+	"""Print the set of supported permissions."""
+	print(_("\nThe following permissions are supported:\n"))
+	fmt = "%-16s %-14s\t%s"
+	print(fmt % (_("NAME"), _("TYPE"), _("NOTES")))
+
+	for (name, note) in sorted(perms_subcmd.iteritems()):
+		print(fmt % (name, _("subcommand"), note))
+
+	for (name, note) in sorted(perms_other.iteritems()):
+		print(fmt % (name, _("other"), note))
+
+	for (name, prop) in sorted(zfs.dataset.proptable.iteritems()):
+		if prop.visible and prop.delegatable():
+			print(fmt % (name, _("property"), ""))
+
+def do_allow():
+	"""Implements the "zfs allow" and "zfs unallow" subcommands."""
+	un = (sys.argv[1] == "unallow")
+
+	def usage(msg=None):
+		parser.print_help()
+		print_perms()
+		if msg:
+			print
+			parser.exit("zfs: error: " + msg)
+		else:
+			parser.exit()
+
+	if un:
+		u = _("""unallow [-rldug] <"everyone"|user|group>[,...]
+	    [<perm|@setname>[,...]] <filesystem|volume>
+	unallow [-rld] -e [<perm|@setname>[,...]] <filesystem|volume>
+	unallow [-r] -c [<perm|@setname>[,...]] <filesystem|volume>
+	unallow [-r] -s @setname [<perm|@setname>[,...]] <filesystem|volume>""")
+		verb = _("remove")
+		sstr = _("undefine permission set")
+	else:
+		u = _("""allow <filesystem|volume>
+	allow [-ldug] <"everyone"|user|group>[,...] <perm|@setname>[,...]
+	    <filesystem|volume>
+	allow [-ld] -e <perm|@setname>[,...] <filesystem|volume>
+	allow -c <perm|@setname>[,...] <filesystem|volume>
+	allow -s @setname <perm|@setname>[,...] <filesystem|volume>""")
+		verb = _("set")
+		sstr = _("define permission set")
+
+	parser = optparse.OptionParser(usage=u, prog="zfs")
+
+	parser.add_option("-l", action="store_true", dest="local",
+	    help=_("%s permission locally") % verb)
+	parser.add_option("-d", action="store_true", dest="descend",
+	    help=_("%s permission for descendents") % verb)
+	parser.add_option("-u", action="store_true", dest="user",
+	    help=_("%s permission for user") % verb)
+	parser.add_option("-g", action="store_true", dest="group",
+	    help=_("%s permission for group") % verb)
+	parser.add_option("-e", action="store_true", dest="everyone",
+	    help=_("%s permission for everyone") % verb)
+	parser.add_option("-c", action="store_true", dest="create",
+	    help=_("%s create time permissions") % verb)
+	parser.add_option("-s", action="store_true", dest="set", help=sstr)
+	if un:
+		parser.add_option("-r", action="store_true", dest="recursive",
+		    help=_("remove permissions recursively"))
+
+	if len(sys.argv) == 3 and not un:
+		# just print the permissions on this fs
+
+		if sys.argv[2] == "-h":
+			# hack to make "zfs allow -h" work
+			usage()
+		ds = zfs.dataset.Dataset(sys.argv[2], snaps=False)
+
+		p = dict()
+		for (fs, raw) in ds.get_fsacl().items():
+			p[fs] = FSPerms(raw)
+
+		for fs in sorted(p.keys(), reverse=True):
+			s = _("---- Permissions on %s ") % fs
+			print(s + "-" * (70-len(s)))
+			print(p[fs])
+		return
+	
+
+	(options, args) = parser.parse_args(sys.argv[2:])
+
+	if sum((bool(options.everyone), bool(options.user),
+	    bool(options.group))) > 1:
+		parser.error(_("-u, -g, and -e are mutually exclusive"))
+
+	def mungeargs(expected_len):
+		if un and len(args) == expected_len-1:
+			return (None, args[expected_len-2])
+		elif len(args) == expected_len:
+			return (args[expected_len-2].split(","),
+			    args[expected_len-1])
+		else:
+			usage(_("wrong number of parameters"))
+
+	if options.set:
+		if options.local or options.descend or options.user or \
+		    options.group or options.everyone or options.create:
+			parser.error(_("invalid option combined with -s"))
+		if args[0][0] != "@":
+			parser.error(_("invalid set name: missing '@' prefix"))
+
+		(perms, fsname) = mungeargs(3)
+		who = args[0]
+	elif options.create:
+		if options.local or options.descend or options.user or \
+		    options.group or options.everyone or options.set:
+			parser.error(_("invalid option combined with -c"))
+
+		(perms, fsname) = mungeargs(2)
+		who = None
+	elif options.everyone:
+		if options.user or options.group or \
+		    options.create or options.set:
+			parser.error(_("invalid option combined with -e"))
+
+		(perms, fsname) = mungeargs(2)
+		who = ["everyone"]
+	else:
+		(perms, fsname) = mungeargs(3)
+		who = args[0].split(",")
+
+	if not options.local and not options.descend:
+		options.local = True
+		options.descend = True
+
+	d = args_to_perms(parser, options, who, perms)
+
+	ds = zfs.dataset.Dataset(fsname, snaps=False)
+
+	if not un and perms:
+		for p in perms:
+			if p[0] == "@" and not hasset(ds, p):
+				parser.error(_("set %s is not defined") % p)
+
+	ds.set_fsacl(un, d)
+	if un and options.recursive:
+		for child in ds.descendents():
+			child.set_fsacl(un, d)

lib/pyzfs/common/dataset.py

@@ -0,0 +1,234 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""Implements the Dataset class, providing methods for manipulating ZFS
+datasets.  Also implements the Property class, which describes ZFS
+properties."""
+
+import zfs.ioctl
+import zfs.util
+import errno
+
+_ = zfs.util._
+
+class Property(object):
+	"""This class represents a ZFS property.  It contains
+	information about the property -- if it's readonly, a number vs
+	string vs index, etc.  Only native properties are represented by
+	this class -- not user properties (eg "user:prop") or userspace
+	properties (eg "userquota@joe")."""
+
+	__slots__ = "name", "number", "type", "default", "attr", "validtypes", \
+	    "values", "colname", "rightalign", "visible", "indextable"
+	__repr__ = zfs.util.default_repr
+
+	def __init__(self, t):
+		"""t is the tuple of information about this property
+		from zfs.ioctl.get_proptable, which should match the
+		members of zprop_desc_t (see zfs_prop.h)."""
+
+		self.name = t[0]
+		self.number = t[1]
+		self.type = t[2]
+		if self.type == "string":
+			self.default = t[3]
+		else:
+			self.default = t[4]
+		self.attr = t[5]
+		self.validtypes = t[6]
+		self.values = t[7]
+		self.colname = t[8]
+		self.rightalign = t[9]
+		self.visible = t[10]
+		self.indextable = t[11]
+
+	def delegatable(self):
+		"""Return True if this property can be delegated with
+		"zfs allow"."""
+		return self.attr != "readonly"
+
+proptable = dict()
+for name, t in zfs.ioctl.get_proptable().iteritems():
+	proptable[name] = Property(t)
+del name, t
+
+def getpropobj(name):
+	"""Return the Property object that is identified by the given
+	name string.  It can be the full name, or the column name."""
+	try:
+		return proptable[name]
+	except KeyError:
+		for p in proptable.itervalues():
+			if p.colname and p.colname.lower() == name:
+				return p
+		raise
+
+class Dataset(object):
+	"""Represents a ZFS dataset (filesystem, snapshot, zvol, clone, etc).
+
+	Generally, this class provides interfaces to the C functions in
+	zfs.ioctl which actually interface with the kernel to manipulate
+	datasets.
+	
+	Unless otherwise noted, any method can raise a ZFSError to
+	indicate failure."""
+
+	__slots__ = "name", "__props"
+	__repr__ = zfs.util.default_repr
+
+	def __init__(self, name, props=None,
+	    types=("filesystem", "volume"), snaps=True):
+		"""Open the named dataset, checking that it exists and
+		is of the specified type.
+		
+		name is the string name of this dataset.
+
+		props is the property settings dict from zfs.ioctl.next_dataset.
+
+		types is an iterable of strings specifying which types
+		of datasets are permitted.  Accepted strings are
+		"filesystem" and "volume".  Defaults to accepting all
+		types.
+
+		snaps is a boolean specifying if snapshots are acceptable.
+
+		Raises a ZFSError if the dataset can't be accessed (eg
+		doesn't exist) or is not of the specified type.
+		"""
+
+		self.name = name
+
+		e = zfs.util.ZFSError(errno.EINVAL,
+		    _("cannot open %s") % name,
+		    _("operation not applicable to datasets of this type"))
+		if "@" in name and not snaps:
+			raise e
+		if not props:
+			props = zfs.ioctl.dataset_props(name)
+		self.__props = props
+		if "volume" not in types and self.getprop("type") == 3:
+			raise e
+		if "filesystem" not in types and self.getprop("type") == 2:
+			raise e
+
+	def getprop(self, propname):
+		"""Return the value of the given property for this dataset.
+
+		Currently only works for native properties (those with a
+		Property object.)
+		
+		Raises KeyError if propname does not specify a native property.
+		Does not raise ZFSError.
+		"""
+
+		p = getpropobj(propname)
+		try:
+			return self.__props[p.name]["value"]
+		except KeyError:
+			return p.default
+
+	def parent(self):
+		"""Return a Dataset representing the parent of this one."""
+		return Dataset(self.name[:self.name.rindex("/")])
+
+	def descendents(self):
+		"""A generator function which iterates over all
+		descendent Datasets (not including snapshots."""
+
+		cookie = 0
+		while True:
+			# next_dataset raises StopIteration when done
+			(name, cookie, props) = \
+			    zfs.ioctl.next_dataset(self.name, False, cookie)
+			ds = Dataset(name, props)
+			yield ds
+			for child in ds.descendents():
+				yield child
+	
+	def userspace(self, prop):
+		"""A generator function which iterates over a
+		userspace-type property.
+
+		prop specifies which property ("userused@",
+		"userquota@", "groupused@", or "groupquota@").
+
+		returns 3-tuple of domain (string), rid (int), and space (int).
+		"""
+
+		d = zfs.ioctl.userspace_many(self.name, prop)
+		for ((domain, rid), space) in d.iteritems():
+			yield (domain, rid, space)
+
+	def userspace_upgrade(self):
+		"""Initialize the accounting information for
+		userused@... and groupused@... properties."""
+		return zfs.ioctl.userspace_upgrade(self.name)
+	
+	def set_fsacl(self, un, d):
+		"""Add to the "zfs allow"-ed permissions on this Dataset.
+
+		un is True if the specified permissions should be removed.
+
+		d is a dict specifying which permissions to add/remove:
+		{ "whostr" -> None # remove all perms for this entity
+		  "whostr" -> { "perm" -> None} # add/remove these perms
+		} """
+		return zfs.ioctl.set_fsacl(self.name, un, d)
+
+	def get_fsacl(self):
+		"""Get the "zfs allow"-ed permissions on the Dataset.
+
+		Return a dict("whostr": { "perm" -> None })."""
+
+		return zfs.ioctl.get_fsacl(self.name)
+
+	def get_holds(self):
+		"""Get the user holds on this Dataset.
+
+		Return a dict("tag": timestamp)."""
+
+		return zfs.ioctl.get_holds(self.name)
+
+def snapshots_fromcmdline(dsnames, recursive):
+	for dsname in dsnames:
+		if not "@" in dsname:
+			raise zfs.util.ZFSError(errno.EINVAL,
+			    _("cannot open %s") % dsname,
+			    _("operation only applies to snapshots"))
+		try:
+			ds = Dataset(dsname)
+			yield ds
+		except zfs.util.ZFSError, e:
+			if not recursive or e.errno != errno.ENOENT:
+				raise
+		if recursive:
+			(base, snapname) = dsname.split('@')
+			parent = Dataset(base)
+			for child in parent.descendents():
+				try:
+					yield Dataset(child.name + "@" +
+					    snapname)
+				except zfs.util.ZFSError, e:
+					if e.errno != errno.ENOENT:
+						raise

lib/pyzfs/common/groupspace.py

@@ -0,0 +1,28 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+import zfs.userspace
+
+do_groupspace = zfs.userspace.do_userspace
+

lib/pyzfs/common/holds.py

@@ -0,0 +1,75 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""This module implements the "zfs holds" subcommand.
+The only public interface is the zfs.holds.do_holds() function."""
+
+import optparse
+import sys
+import errno
+import time
+import zfs.util
+import zfs.dataset
+import zfs.table
+
+_ = zfs.util._
+
+def do_holds():
+	"""Implements the "zfs holds" subcommand."""
+	def usage(msg=None):
+		parser.print_help()
+		if msg:
+			print
+			parser.exit("zfs: error: " + msg)
+		else:
+			parser.exit()
+
+	u = _("""holds [-r] <snapshot> ...""")
+
+	parser = optparse.OptionParser(usage=u, prog="zfs")
+
+	parser.add_option("-r", action="store_true", dest="recursive",
+	    help=_("list holds recursively"))
+
+	(options, args) = parser.parse_args(sys.argv[2:])
+
+	if len(args) < 1:
+		usage(_("missing snapshot argument"))
+
+	fields = ("name", "tag", "timestamp")
+	rjustfields = ()
+	printing = False 
+	gotone = False
+	t = zfs.table.Table(fields, rjustfields) 
+	for ds in zfs.dataset.snapshots_fromcmdline(args, options.recursive):
+		gotone = True
+		for tag, tm in ds.get_holds().iteritems():
+			val = {"name": ds.name, "tag": tag,
+			    "timestamp": time.ctime(tm)}
+			t.addline(ds.name, val)
+			printing = True
+	if printing:
+		t.printme()
+	elif not gotone:
+		raise zfs.util.ZFSError(errno.ENOENT, _("no matching datasets"))

lib/pyzfs/common/ioctl.c

@@ -0,0 +1,543 @@
+/*
+ * 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 2010 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <Python.h>
+#include <sys/zfs_ioctl.h>
+#include <sys/fs/zfs.h>
+#include <strings.h>
+#include <unistd.h>
+#include <libnvpair.h>
+#include <libintl.h>
+#include <libzfs.h>
+#include "zfs_prop.h"
+
+static PyObject *ZFSError;
+static int zfsdevfd;
+
+#ifdef __lint
+#define	dgettext(x, y) y
+#endif
+
+#define	_(s) dgettext(TEXT_DOMAIN, s)
+
+/*PRINTFLIKE1*/
+static void
+seterr(char *fmt, ...)
+{
+	char errstr[1024];
+	va_list v;
+
+	va_start(v, fmt);
+	(void) vsnprintf(errstr, sizeof (errstr), fmt, v);
+	va_end(v);
+
+	PyErr_SetObject(ZFSError, Py_BuildValue("is", errno, errstr));
+}
+
+static char cmdstr[HIS_MAX_RECORD_LEN];
+
+static int
+ioctl_with_cmdstr(int ioc, zfs_cmd_t *zc)
+{
+	int err;
+
+	if (cmdstr[0])
+		zc->zc_history = (uint64_t)(uintptr_t)cmdstr;
+	err = ioctl(zfsdevfd, ioc, zc);
+	cmdstr[0] = '\0';
+	return (err);
+}
+
+static PyObject *
+nvl2py(nvlist_t *nvl)
+{
+	PyObject *pyo;
+	nvpair_t *nvp;
+
+	pyo = PyDict_New();
+
+	for (nvp = nvlist_next_nvpair(nvl, NULL); nvp;
+	    nvp = nvlist_next_nvpair(nvl, nvp)) {
+		PyObject *pyval;
+		char *sval;
+		uint64_t ival;
+		boolean_t bval;
+		nvlist_t *nval;
+
+		switch (nvpair_type(nvp)) {
+		case DATA_TYPE_STRING:
+			(void) nvpair_value_string(nvp, &sval);
+			pyval = Py_BuildValue("s", sval);
+			break;
+
+		case DATA_TYPE_UINT64:
+			(void) nvpair_value_uint64(nvp, &ival);
+			pyval = Py_BuildValue("K", ival);
+			break;
+
+		case DATA_TYPE_NVLIST:
+			(void) nvpair_value_nvlist(nvp, &nval);
+			pyval = nvl2py(nval);
+			break;
+
+		case DATA_TYPE_BOOLEAN:
+			Py_INCREF(Py_None);
+			pyval = Py_None;
+			break;
+
+		case DATA_TYPE_BOOLEAN_VALUE:
+			(void) nvpair_value_boolean_value(nvp, &bval);
+			pyval = Py_BuildValue("i", bval);
+			break;
+
+		default:
+			PyErr_SetNone(PyExc_ValueError);
+			Py_DECREF(pyo);
+			return (NULL);
+		}
+
+		PyDict_SetItemString(pyo, nvpair_name(nvp), pyval);
+		Py_DECREF(pyval);
+	}
+
+	return (pyo);
+}
+
+static nvlist_t *
+dict2nvl(PyObject *d)
+{
+	nvlist_t *nvl;
+	int err;
+	PyObject *key, *value;
+	int pos = 0;
+
+	if (!PyDict_Check(d)) {
+		PyErr_SetObject(PyExc_ValueError, d);
+		return (NULL);
+	}
+
+	err = nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0);
+	assert(err == 0);
+
+	while (PyDict_Next(d, &pos, &key, &value)) {
+		char *keystr = PyString_AsString(key);
+		if (keystr == NULL) {
+			PyErr_SetObject(PyExc_KeyError, key);
+			nvlist_free(nvl);
+			return (NULL);
+		}
+
+		if (PyDict_Check(value)) {
+			nvlist_t *valnvl = dict2nvl(value);
+			err = nvlist_add_nvlist(nvl, keystr, valnvl);
+			nvlist_free(valnvl);
+		} else if (value == Py_None) {
+			err = nvlist_add_boolean(nvl, keystr);
+		} else if (PyString_Check(value)) {
+			char *valstr = PyString_AsString(value);
+			err = nvlist_add_string(nvl, keystr, valstr);
+		} else if (PyInt_Check(value)) {
+			uint64_t valint = PyInt_AsUnsignedLongLongMask(value);
+			err = nvlist_add_uint64(nvl, keystr, valint);
+		} else if (PyBool_Check(value)) {
+			boolean_t valbool = value == Py_True ? B_TRUE : B_FALSE;
+			err = nvlist_add_boolean_value(nvl, keystr, valbool);
+		} else {
+			PyErr_SetObject(PyExc_ValueError, value);
+			nvlist_free(nvl);
+			return (NULL);
+		}
+		assert(err == 0);
+	}
+
+	return (nvl);
+}
+
+static PyObject *
+fakepropval(uint64_t value)
+{
+	PyObject *d = PyDict_New();
+	PyDict_SetItemString(d, "value", Py_BuildValue("K", value));
+	return (d);
+}
+
+static void
+add_ds_props(zfs_cmd_t *zc, PyObject *nvl)
+{
+	dmu_objset_stats_t *s = &zc->zc_objset_stats;
+	PyDict_SetItemString(nvl, "numclones",
+	    fakepropval(s->dds_num_clones));
+	PyDict_SetItemString(nvl, "issnap",
+	    fakepropval(s->dds_is_snapshot));
+	PyDict_SetItemString(nvl, "inconsistent",
+	    fakepropval(s->dds_inconsistent));
+}
+
+/* On error, returns NULL but does not set python exception. */
+static PyObject *
+ioctl_with_dstnv(int ioc, zfs_cmd_t *zc)
+{
+	int nvsz = 2048;
+	void *nvbuf;
+	PyObject *pynv = NULL;
+
+again:
+	nvbuf = malloc(nvsz);
+	zc->zc_nvlist_dst_size = nvsz;
+	zc->zc_nvlist_dst = (uintptr_t)nvbuf;
+
+	if (ioctl(zfsdevfd, ioc, zc) == 0) {
+		nvlist_t *nvl;
+
+		errno = nvlist_unpack(nvbuf, zc->zc_nvlist_dst_size, &nvl, 0);
+		if (errno == 0) {
+			pynv = nvl2py(nvl);
+			nvlist_free(nvl);
+		}
+	} else if (errno == ENOMEM) {
+		free(nvbuf);
+		nvsz = zc->zc_nvlist_dst_size;
+		goto again;
+	}
+	free(nvbuf);
+	return (pynv);
+}
+
+static PyObject *
+py_next_dataset(PyObject *self, PyObject *args)
+{
+	int ioc;
+	uint64_t cookie;
+	zfs_cmd_t zc = { 0 };
+	int snaps;
+	char *name;
+	PyObject *nvl;
+	PyObject *ret = NULL;
+
+	if (!PyArg_ParseTuple(args, "siK", &name, &snaps, &cookie))
+		return (NULL);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+	zc.zc_cookie = cookie;
+
+	if (snaps)
+		ioc = ZFS_IOC_SNAPSHOT_LIST_NEXT;
+	else
+		ioc = ZFS_IOC_DATASET_LIST_NEXT;
+
+	nvl = ioctl_with_dstnv(ioc, &zc);
+	if (nvl) {
+		add_ds_props(&zc, nvl);
+		ret = Py_BuildValue("sKO", zc.zc_name, zc.zc_cookie, nvl);
+		Py_DECREF(nvl);
+	} else if (errno == ESRCH) {
+		PyErr_SetNone(PyExc_StopIteration);
+	} else {
+		if (snaps)
+			seterr(_("cannot get snapshots of %s"), name);
+		else
+			seterr(_("cannot get child datasets of %s"), name);
+	}
+	return (ret);
+}
+
+static PyObject *
+py_dataset_props(PyObject *self, PyObject *args)
+{
+	zfs_cmd_t zc = { 0 };
+	int snaps;
+	char *name;
+	PyObject *nvl;
+
+	if (!PyArg_ParseTuple(args, "s", &name))
+		return (NULL);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+
+	nvl = ioctl_with_dstnv(ZFS_IOC_OBJSET_STATS, &zc);
+	if (nvl) {
+		add_ds_props(&zc, nvl);
+	} else {
+		seterr(_("cannot access dataset %s"), name);
+	}
+	return (nvl);
+}
+
+static PyObject *
+py_get_fsacl(PyObject *self, PyObject *args)
+{
+	zfs_cmd_t zc = { 0 };
+	char *name;
+	PyObject *nvl;
+
+	if (!PyArg_ParseTuple(args, "s", &name))
+		return (NULL);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+
+	nvl = ioctl_with_dstnv(ZFS_IOC_GET_FSACL, &zc);
+	if (nvl == NULL)
+		seterr(_("cannot get permissions on %s"), name);
+
+	return (nvl);
+}
+
+static PyObject *
+py_set_fsacl(PyObject *self, PyObject *args)
+{
+	int un;
+	size_t nvsz;
+	zfs_cmd_t zc = { 0 };
+	char *name, *nvbuf;
+	PyObject *dict, *file;
+	nvlist_t *nvl;
+	int err;
+
+	if (!PyArg_ParseTuple(args, "siO!", &name, &un,
+	    &PyDict_Type, &dict))
+		return (NULL);
+
+	nvl = dict2nvl(dict);
+	if (nvl == NULL)
+		return (NULL);
+
+	err = nvlist_size(nvl, &nvsz, NV_ENCODE_NATIVE);
+	assert(err == 0);
+	nvbuf = malloc(nvsz);
+	err = nvlist_pack(nvl, &nvbuf, &nvsz, NV_ENCODE_NATIVE, 0);
+	assert(err == 0);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+	zc.zc_nvlist_src_size = nvsz;
+	zc.zc_nvlist_src = (uintptr_t)nvbuf;
+	zc.zc_perm_action = un;
+
+	err = ioctl_with_cmdstr(ZFS_IOC_SET_FSACL, &zc);
+	free(nvbuf);
+	if (err) {
+		seterr(_("cannot set permissions on %s"), name);
+		return (NULL);
+	}
+
+	Py_RETURN_NONE;
+}
+
+static PyObject *
+py_get_holds(PyObject *self, PyObject *args)
+{
+	zfs_cmd_t zc = { 0 };
+	char *name;
+	PyObject *nvl;
+
+	if (!PyArg_ParseTuple(args, "s", &name))
+		return (NULL);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+
+	nvl = ioctl_with_dstnv(ZFS_IOC_GET_HOLDS, &zc);
+	if (nvl == NULL)
+		seterr(_("cannot get holds for %s"), name);
+
+	return (nvl);
+}
+
+static PyObject *
+py_userspace_many(PyObject *self, PyObject *args)
+{
+	zfs_cmd_t zc = { 0 };
+	zfs_userquota_prop_t type;
+	char *name, *propname;
+	int bufsz = 1<<20;
+	void *buf;
+	PyObject *dict, *file;
+	int error;
+
+	if (!PyArg_ParseTuple(args, "ss", &name, &propname))
+		return (NULL);
+
+	for (type = 0; type < ZFS_NUM_USERQUOTA_PROPS; type++)
+		if (strcmp(propname, zfs_userquota_prop_prefixes[type]) == 0)
+			break;
+	if (type == ZFS_NUM_USERQUOTA_PROPS) {
+		PyErr_SetString(PyExc_KeyError, propname);
+		return (NULL);
+	}
+
+	dict = PyDict_New();
+	buf = malloc(bufsz);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+	zc.zc_objset_type = type;
+	zc.zc_cookie = 0;
+
+	while (1) {
+		zfs_useracct_t *zua = buf;
+
+		zc.zc_nvlist_dst = (uintptr_t)buf;
+		zc.zc_nvlist_dst_size = bufsz;
+
+		error = ioctl(zfsdevfd, ZFS_IOC_USERSPACE_MANY, &zc);
+		if (error || zc.zc_nvlist_dst_size == 0)
+			break;
+
+		while (zc.zc_nvlist_dst_size > 0) {
+			PyObject *pykey, *pyval;
+
+			pykey = Py_BuildValue("sI",
+			    zua->zu_domain, zua->zu_rid);
+			pyval = Py_BuildValue("K", zua->zu_space);
+			PyDict_SetItem(dict, pykey, pyval);
+			Py_DECREF(pykey);
+			Py_DECREF(pyval);
+
+			zua++;
+			zc.zc_nvlist_dst_size -= sizeof (zfs_useracct_t);
+		}
+	}
+
+	free(buf);
+
+	if (error != 0) {
+		Py_DECREF(dict);
+		seterr(_("cannot get %s property on %s"), propname, name);
+		return (NULL);
+	}
+
+	return (dict);
+}
+
+static PyObject *
+py_userspace_upgrade(PyObject *self, PyObject *args)
+{
+	zfs_cmd_t zc = { 0 };
+	char *name;
+	int error;
+
+	if (!PyArg_ParseTuple(args, "s", &name))
+		return (NULL);
+
+	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
+	error = ioctl(zfsdevfd, ZFS_IOC_USERSPACE_UPGRADE, &zc);
+
+	if (error != 0) {
+		seterr(_("cannot initialize user accounting information on %s"),
+		    name);
+		return (NULL);
+	}
+
+	Py_RETURN_NONE;
+}
+
+static PyObject *
+py_set_cmdstr(PyObject *self, PyObject *args)
+{
+	char *str;
+
+	if (!PyArg_ParseTuple(args, "s", &str))
+		return (NULL);
+
+	(void) strlcpy(cmdstr, str, sizeof (cmdstr));
+
+	Py_RETURN_NONE;
+}
+
+static PyObject *
+py_get_proptable(PyObject *self, PyObject *args)
+{
+	zprop_desc_t *t = zfs_prop_get_table();
+	PyObject *d = PyDict_New();
+	zfs_prop_t i;
+
+	for (i = 0; i < ZFS_NUM_PROPS; i++) {
+		zprop_desc_t *p = &t[i];
+		PyObject *tuple;
+		static const char *typetable[] =
+		    {"number", "string", "index"};
+		static const char *attrtable[] =
+		    {"default", "readonly", "inherit", "onetime"};
+		PyObject *indextable;
+
+		if (p->pd_proptype == PROP_TYPE_INDEX) {
+			const zprop_index_t *it = p->pd_table;
+			indextable = PyDict_New();
+			int j;
+			for (j = 0; it[j].pi_name; j++) {
+				PyDict_SetItemString(indextable,
+				    it[j].pi_name,
+				    Py_BuildValue("K", it[j].pi_value));
+			}
+		} else {
+			Py_INCREF(Py_None);
+			indextable = Py_None;
+		}
+
+		tuple = Py_BuildValue("sissKsissiiO",
+		    p->pd_name, p->pd_propnum, typetable[p->pd_proptype],
+		    p->pd_strdefault, p->pd_numdefault,
+		    attrtable[p->pd_attr], p->pd_types,
+		    p->pd_values, p->pd_colname,
+		    p->pd_rightalign, p->pd_visible, indextable);
+		PyDict_SetItemString(d, p->pd_name, tuple);
+		Py_DECREF(tuple);
+	}
+
+	return (d);
+}
+
+static PyMethodDef zfsmethods[] = {
+	{"next_dataset", py_next_dataset, METH_VARARGS,
+	    "Get next child dataset or snapshot."},
+	{"get_fsacl", py_get_fsacl, METH_VARARGS, "Get allowed permissions."},
+	{"set_fsacl", py_set_fsacl, METH_VARARGS, "Set allowed permissions."},
+	{"userspace_many", py_userspace_many, METH_VARARGS,
+	    "Get user space accounting."},
+	{"userspace_upgrade", py_userspace_upgrade, METH_VARARGS,
+	    "Upgrade fs to enable user space accounting."},
+	{"set_cmdstr", py_set_cmdstr, METH_VARARGS,
+	    "Set command string for history logging."},
+	{"dataset_props", py_dataset_props, METH_VARARGS,
+	    "Get dataset properties."},
+	{"get_proptable", py_get_proptable, METH_NOARGS,
+	    "Get property table."},
+	{"get_holds", py_get_holds, METH_VARARGS, "Get user holds."},
+	{NULL, NULL, 0, NULL}
+};
+
+void
+initioctl(void)
+{
+	PyObject *zfs_ioctl = Py_InitModule("zfs.ioctl", zfsmethods);
+	PyObject *zfs_util = PyImport_ImportModule("zfs.util");
+	PyObject *devfile;
+
+	if (zfs_util == NULL)
+		return;
+
+	ZFSError = PyObject_GetAttrString(zfs_util, "ZFSError");
+	devfile = PyObject_GetAttrString(zfs_util, "dev");
+	zfsdevfd = PyObject_AsFileDescriptor(devfile);
+
+	zfs_prop_init();
+}

lib/pyzfs/common/table.py

@@ -0,0 +1,70 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+import zfs.util
+
+class Table:
+	__slots__ = "fields", "rjustfields", "maxfieldlen", "lines"
+	__repr__ = zfs.util.default_repr
+
+	def __init__(self, fields, rjustfields=()):
+		# XXX maybe have a defaults, too?
+		self.fields = fields
+		self.rjustfields = rjustfields
+		self.maxfieldlen = dict.fromkeys(fields, 0)
+		self.lines = list()
+	
+	def __updatemax(self, k, v):
+		self.maxfieldlen[k] = max(self.maxfieldlen.get(k, None), v)
+
+	def addline(self, sortkey, values):
+		"""values is a dict from field name to value"""
+
+		va = list()
+		for f in self.fields:
+			v = str(values[f])
+			va.append(v)
+			self.__updatemax(f, len(v))
+		self.lines.append((sortkey, va))
+
+	def printme(self, headers=True):
+		if headers:
+			d = dict([(f, f.upper()) for f in self.fields])
+			self.addline(None, d)
+
+		self.lines.sort()
+		for (k, va) in self.lines:
+			line = str()
+			for i in range(len(self.fields)):
+				if not headers:
+					line += va[i]
+					line += "\t"
+				else:
+					if self.fields[i] in self.rjustfields:
+						fmt = "%*s  "
+					else:
+						fmt = "%-*s  "
+					mfl = self.maxfieldlen[self.fields[i]]
+					line += fmt % (mfl, va[i])
+			print(line)

lib/pyzfs/common/unallow.py

@@ -0,0 +1,27 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+import zfs.allow
+
+do_unallow = zfs.allow.do_allow

lib/pyzfs/common/userspace.py

@@ -0,0 +1,246 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""This module implements the "zfs userspace" and "zfs groupspace" subcommands.
+The only public interface is the zfs.userspace.do_userspace() function."""
+
+import optparse
+import sys
+import pwd
+import grp
+import errno
+import solaris.misc
+import zfs.util
+import zfs.ioctl
+import zfs.dataset
+import zfs.table
+
+_ = zfs.util._
+
+# map from property name prefix -> (field name, isgroup)
+props = {
+    "userused@": ("used", False),
+    "userquota@": ("quota", False),
+    "groupused@": ("used", True),
+    "groupquota@": ("quota", True),
+}
+
+def skiptype(options, prop):
+	"""Return True if this property (eg "userquota@") should be skipped."""
+	(field, isgroup) = props[prop]
+	if field not in options.fields:
+		return True
+	if isgroup and "posixgroup" not in options.types and \
+	    "smbgroup" not in options.types:
+		return True
+	if not isgroup and "posixuser" not in options.types and \
+	    "smbuser" not in options.types:
+		return True
+	return False
+
+def new_entry(options, isgroup, domain, rid):
+	"""Return a dict("field": value) for this domain (string) + rid (int)"""
+
+	if domain:
+		idstr = "%s-%u" % (domain, rid)
+	else:
+		idstr = "%u" % rid
+
+	(typename, mapfunc) = {
+	    (1, 1): ("SMB Group",   lambda id: solaris.misc.sid_to_name(id, 0)),
+	    (1, 0): ("POSIX Group", lambda id: grp.getgrgid(int(id)).gr_name),
+	    (0, 1): ("SMB User",    lambda id: solaris.misc.sid_to_name(id, 1)),
+	    (0, 0): ("POSIX User",  lambda id: pwd.getpwuid(int(id)).pw_name)
+	}[isgroup, bool(domain)]
+
+	if typename.lower().replace(" ", "") not in options.types:
+		return None
+
+	v = dict()
+	v["type"] = typename
+
+	# python's getpwuid/getgrgid is confused by ephemeral uids
+	if not options.noname and rid < 1<<31:
+		try:
+			v["name"] = mapfunc(idstr)
+		except KeyError:
+			pass
+
+	if "name" not in v:
+		v["name"] = idstr
+		if not domain:
+			# it's just a number, so pad it with spaces so
+			# that it will sort numerically
+			v["name.sort"] = "%20d" % rid
+	# fill in default values
+	v["used"] = "0"
+	v["used.sort"] = 0
+	v["quota"] = "none"
+	v["quota.sort"] = 0
+	return v
+
+def process_one_raw(acct, options, prop, elem):
+	"""Update the acct dict to incorporate the
+	information from this elem from Dataset.userspace(prop)."""
+
+	(domain, rid, value) = elem
+	(field, isgroup) = props[prop]
+
+	if options.translate and domain:
+		try:
+			rid = solaris.misc.sid_to_id("%s-%u" % (domain, rid),
+			    not isgroup)
+			domain = None
+		except KeyError:
+			pass;
+	key = (isgroup, domain, rid)
+		
+	try:
+		v = acct[key]
+	except KeyError:
+		v = new_entry(options, isgroup, domain, rid)
+		if not v:
+			return
+		acct[key] = v
+
+	# Add our value to an existing value, which may be present if
+	# options.translate is set.
+	value = v[field + ".sort"] = value + v[field + ".sort"]
+
+	if options.parsable:
+		v[field] = str(value)
+	else:
+		v[field] = zfs.util.nicenum(value)
+
+def do_userspace():
+	"""Implements the "zfs userspace" and "zfs groupspace" subcommands."""
+
+	def usage(msg=None):
+		parser.print_help()
+		if msg:
+			print
+			parser.exit("zfs: error: " + msg)
+		else:
+			parser.exit()
+
+	if sys.argv[1] == "userspace":
+		defaulttypes = "posixuser,smbuser"
+	else:
+		defaulttypes = "posixgroup,smbgroup"
+
+	fields = ("type", "name", "used", "quota")
+	rjustfields = ("used", "quota")
+	types = ("all", "posixuser", "smbuser", "posixgroup", "smbgroup")
+
+	u = _("%s [-niHp] [-o field[,...]] [-sS field] ... \n") % sys.argv[1]
+	u += _("    [-t type[,...]] <filesystem|snapshot>")
+	parser = optparse.OptionParser(usage=u, prog="zfs")
+
+	parser.add_option("-n", action="store_true", dest="noname",
+	    help=_("Print numeric ID instead of user/group name"))
+	parser.add_option("-i", action="store_true", dest="translate",
+	    help=_("translate SID to posix (possibly ephemeral) ID"))
+	parser.add_option("-H", action="store_true", dest="noheaders",
+	    help=_("no headers, tab delimited output"))
+	parser.add_option("-p", action="store_true", dest="parsable",
+	    help=_("exact (parsable) numeric output"))
+	parser.add_option("-o", dest="fields", metavar="field[,...]",
+	    default="type,name,used,quota",
+	    help=_("print only these fields (eg type,name,used,quota)"))
+	parser.add_option("-s", dest="sortfields", metavar="field",
+	    type="choice", choices=fields, default=list(),
+	    action="callback", callback=zfs.util.append_with_opt,
+	    help=_("sort field"))
+	parser.add_option("-S", dest="sortfields", metavar="field",
+	    type="choice", choices=fields, #-s sets the default
+	    action="callback", callback=zfs.util.append_with_opt,
+	    help=_("reverse sort field"))
+	parser.add_option("-t", dest="types", metavar="type[,...]",
+	    default=defaulttypes,
+	    help=_("print only these types (eg posixuser,smbuser,posixgroup,smbgroup,all)"))
+
+	(options, args) = parser.parse_args(sys.argv[2:])
+	if len(args) != 1:
+		usage(_("wrong number of arguments"))
+	dsname = args[0]
+
+	options.fields = options.fields.split(",")
+	for f in options.fields:
+		if f not in fields:
+			usage(_("invalid field %s") % f)
+
+	options.types = options.types.split(",")
+	for t in options.types:
+		if t not in types:
+			usage(_("invalid type %s") % t)
+
+	if not options.sortfields:
+		options.sortfields = [("-s", "type"), ("-s", "name")]
+
+	if "all" in options.types:
+		options.types = types[1:]
+
+	ds = zfs.dataset.Dataset(dsname, types=("filesystem"))
+
+	if ds.getprop("zoned") and solaris.misc.isglobalzone():
+		options.noname = True
+
+	if not ds.getprop("useraccounting"):
+		print(_("Initializing accounting information on old filesystem, please wait..."))
+		ds.userspace_upgrade()
+
+	# gather and process accounting information
+	# Due to -i, we need to keep a dict, so we can potentially add
+	# together the posix ID and SID's usage.  Grr.
+	acct = dict()
+	for prop in props.keys():
+		if skiptype(options, prop):
+			continue;
+		for elem in ds.userspace(prop):
+			process_one_raw(acct, options, prop, elem)
+
+	def cmpkey(val):
+		l = list()
+		for (opt, field) in options.sortfields:
+			try:
+				n = val[field + ".sort"]
+			except KeyError:
+				n = val[field]
+			if opt == "-S":
+				# reverse sorting
+				try:
+					n = -n
+				except TypeError:
+					# it's a string; decompose it
+					# into an array of integers,
+					# each one the negative of that
+					# character
+					n = [-ord(c) for c in n]
+			l.append(n)
+		return l
+
+	t = zfs.table.Table(options.fields, rjustfields)
+	for val in acct.itervalues():
+		t.addline(cmpkey(val), val)
+	t.printme(not options.noheaders)

lib/pyzfs/common/util.py

@@ -0,0 +1,141 @@
+#! /usr/bin/python2.6
+#
+# 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 (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
+#
+
+"""This module provides utility functions for ZFS.
+zfs.util.dev -- a file object of /dev/zfs """
+
+import gettext
+import errno
+import os
+import solaris.misc
+# Note: this module (zfs.util) should not import zfs.ioctl, because that
+# would introduce a circular dependency
+
+errno.ECANCELED = 47
+errno.ENOTSUP = 48
+
+dev = open("/dev/zfs", "w")
+
+try:
+	_ = gettext.translation("SUNW_OST_OSLIB", "/usr/lib/locale",
+	    fallback=True).gettext
+except:
+	_ = solaris.misc.gettext
+
+def default_repr(self):
+	"""A simple __repr__ function."""
+	if self.__slots__:
+		str = "<" + self.__class__.__name__
+		for v in self.__slots__:
+			str += " %s: %r" % (v, getattr(self, v))
+		return str + ">"
+	else:
+		return "<%s %s>" % \
+		    (self.__class__.__name__, repr(self.__dict__))
+
+class ZFSError(StandardError):
+	"""This exception class represents a potentially user-visible
+	ZFS error.  If uncaught, it will be printed and the process will
+	exit with exit code 1.
+	
+	errno -- the error number (eg, from ioctl(2))."""
+
+	__slots__ = "why", "task", "errno"
+	__repr__ = default_repr
+
+	def __init__(self, eno, task=None, why=None):
+		"""Create a ZFS exception.
+		eno -- the error number (errno)
+		task -- a string describing the task that failed
+		why -- a string describing why it failed (defaults to
+		    strerror(eno))"""
+
+		self.errno = eno
+		self.task = task
+		self.why = why
+
+	def __str__(self):
+		s = ""
+		if self.task:
+			s += self.task + ": "
+		if self.why:
+			s += self.why
+		else:
+			s += self.strerror
+		return s
+
+	__strs = {
+		errno.EPERM: _("permission denied"),
+		errno.ECANCELED:
+		    _("delegated administration is disabled on pool"),
+		errno.EINTR: _("signal received"),
+		errno.EIO: _("I/O error"),
+		errno.ENOENT: _("dataset does not exist"),
+		errno.ENOSPC: _("out of space"),
+		errno.EEXIST: _("dataset already exists"),
+		errno.EBUSY: _("dataset is busy"),
+		errno.EROFS:
+		    _("snapshot permissions cannot be modified"),
+		errno.ENAMETOOLONG: _("dataset name is too long"),
+		errno.ENOTSUP: _("unsupported version"),
+		errno.EAGAIN: _("pool I/O is currently suspended"),
+	}
+
+	__strs[errno.EACCES] = __strs[errno.EPERM]
+	__strs[errno.ENXIO] = __strs[errno.EIO]
+	__strs[errno.ENODEV] = __strs[errno.EIO]
+	__strs[errno.EDQUOT] = __strs[errno.ENOSPC]
+
+	@property
+	def strerror(self):
+		return ZFSError.__strs.get(self.errno, os.strerror(self.errno))
+
+def nicenum(num):
+	"""Return a nice string (eg "1.23M") for this integer."""
+	index = 0;
+	n = num;
+
+	while n >= 1024:
+		n /= 1024
+		index += 1
+
+	u = " KMGTPE"[index]
+	if index == 0:
+		return "%u" % n;
+	elif n >= 100 or num & ((1024*index)-1) == 0:
+		# it's an exact multiple of its index, or it wouldn't
+		# fit as floating point, so print as an integer
+		return "%u%c" % (n, u)
+	else:
+		# due to rounding, it's tricky to tell what precision to
+		# use; try each precision and see which one fits
+		for i in (2, 1, 0):
+			s = "%.*f%c" % (i, float(num) / (1<<(10*index)), u)
+			if len(s) <= 5:
+				return s
+
+def append_with_opt(option, opt, value, parser):
+	"""A function for OptionParser which appends a tuple (opt, value)."""
+	getattr(parser.values, option.dest).append((opt, value))
+