Merge remote-tracking branch 'origin/stable/10' into devel

19 files changed, 373 insertions, 73 deletions

diff --git a/sbin/bsdlabel/bsdlabel.8 b/sbin/bsdlabel/bsdlabel.8
index d31187b..8137171 100644
--- a/sbin/bsdlabel/bsdlabel.8
+++ b/sbin/bsdlabel/bsdlabel.8
@@ -445,7 +445,10 @@ to properly
 recognize the disk:
 .Bd -literal -offset indent
 dd if=/dev/zero of=/dev/da0 bs=512 count=32
-fdisk -BI da0
+gpart create -s MBR da0
+gpart add -t freebsd da0
+gpart set -a active -i 1 da0
+gpart bootcode -b /boot/mbr da0
 dd if=/dev/zero of=/dev/da0s1 bs=512 count=32
 bsdlabel -w -B da0s1
 bsdlabel -e da0s1
@@ -495,6 +498,5 @@ are not generally compatible.
 .Xr md 4 ,
 .Xr disktab 5 ,
 .Xr boot0cfg 8 ,
-.Xr fdisk 8 ,
 .Xr gpart 8 ,
 .Xr newfs 8
diff --git a/sbin/devd/devd.cc b/sbin/devd/devd.cc
index a0cb037..1667219 100644
--- a/sbin/devd/devd.cc
+++ b/sbin/devd/devd.cc
@@ -108,15 +108,26 @@ __FBSDID("$FreeBSD$");
 /*
  * Since the client socket is nonblocking, we must increase its send buffer to
  * handle brief event storms.  On FreeBSD, AF_UNIX sockets don't have a receive
- * buffer, so the client can't increate the buffersize by itself.
+ * buffer, so the client can't increase the buffersize by itself.
  *
  * For example, when creating a ZFS pool, devd emits one 165 character
- * resource.fs.zfs.statechange message for each vdev in the pool.  A 64k
- * buffer has enough space for almost 400 drives, which would be very large but
- * not impossibly large pool.  A 128k buffer has enough space for 794 drives,
- * which is more than can fit in a rack with modern technology.
+ * resource.fs.zfs.statechange message for each vdev in the pool.  The kernel
+ * allocates a 4608B mbuf for each message.  Modern technology places a limit of
+ * roughly 450 drives/rack, and it's unlikely that a zpool will ever be larger
+ * than that.
+ *
+ * 450 drives * 165 bytes / drive = 74250B of data in the sockbuf
+ * 450 drives * 4608B / drive = 2073600B of mbufs in the sockbuf
+ *
+ * We can't directly set the sockbuf's mbuf limit, but we can do it indirectly.
+ * The kernel sets it to the minimum of a hard-coded maximum value and sbcc *
+ * kern.ipc.sockbuf_waste_factor, where sbcc is the socket buffer size set by
+ * the user.  The default value of kern.ipc.sockbuf_waste_factor is 8.  If we
+ * set the bufsize to 256k and use the kern.ipc.sockbuf_waste_factor, then the
+ * kernel will set the mbuf limit to 2MB, which is just large enough for 450
+ * drives.  It also happens to be the same as the hardcoded maximum value.
  */
-#define CLIENT_BUFSIZE 131072
+#define CLIENT_BUFSIZE 262144
 
 using namespace std;
 
diff --git a/sbin/dumpfs/dumpfs.8 b/sbin/dumpfs/dumpfs.8
index feb8758..0d41059 100644
--- a/sbin/dumpfs/dumpfs.8
+++ b/sbin/dumpfs/dumpfs.8
@@ -98,10 +98,9 @@ The
 flag is needed if the filesystem uses
 .Xr gjournal 8 .
 .Sh SEE ALSO
-.Xr disktab 5 ,
 .Xr fs 5 ,
-.Xr disklabel 8 ,
 .Xr fsck 8 ,
+.Xr gpart 8 ,
 .Xr newfs 8 ,
 .Xr tunefs 8
 .Sh HISTORY
diff --git a/sbin/fdisk/fdisk.8 b/sbin/fdisk/fdisk.8
index fcab133..6894ab9 100644
--- a/sbin/fdisk/fdisk.8
+++ b/sbin/fdisk/fdisk.8
@@ -39,6 +39,13 @@ The
 utility can be used to divide space on the disk into slices and set one
 active.
 .Sh DESCRIPTION
+.Bf -symbolic
+This command is obsolete.
+Users are advised to use
+.Xr gpart 8
+instead.
+.Ef
+.Pp
 The
 .Fx
 utility,
diff --git a/sbin/ffsinfo/ffsinfo.8 b/sbin/ffsinfo/ffsinfo.8
index 9753cf7..0c114bd 100644
--- a/sbin/ffsinfo/ffsinfo.8
+++ b/sbin/ffsinfo/ffsinfo.8
@@ -121,9 +121,9 @@ to
 .Pa /var/tmp/ffsinfo
 with all available information.
 .Sh SEE ALSO
-.Xr disklabel 8 ,
 .Xr dumpfs 8 ,
 .Xr fsck 8 ,
+.Xr gpart 8 ,
 .Xr growfs 8 ,
 .Xr gvinum 8 ,
 .Xr newfs 8 ,
diff --git a/sbin/geom/class/nop/gnop.8 b/sbin/geom/class/nop/gnop.8
index 0103579..ae350d0 100644
--- a/sbin/geom/class/nop/gnop.8
+++ b/sbin/geom/class/nop/gnop.8
@@ -24,7 +24,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd September 15, 2015
+.Dd October 10, 2015
 .Dt GNOP 8
 .Os
 .Sh NAME
@@ -72,9 +72,10 @@ The
 utility is used for setting up transparent providers on existing ones.
 Its main purpose is testing other GEOM classes, as it allows forced provider
 removal and I/O error simulation with a given probability.
-It also gathers the following statistics: number of read requests, number of
-write requests, number of bytes read and number of bytes written.
-In addition, it can be used as a good starting point for implementing new GEOM
+It also gathers statistics on the number of read, write, delete,
+getattr, flush, and other requests, and the number of bytes read and written.
+.Nm
+can also be used as a good starting point for implementing new GEOM
 classes.
 .Pp
 The first argument to
diff --git a/sbin/geom/class/part/gpart.8 b/sbin/geom/class/part/gpart.8
index 3421ba0..ba579bf 100644
--- a/sbin/geom/class/part/gpart.8
+++ b/sbin/geom/class/part/gpart.8
@@ -24,7 +24,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd July 14, 2015
+.Dd December 10, 2015
 .Dt GPART 8
 .Os
 .Sh NAME
@@ -1107,15 +1107,6 @@ and
 .Cm recover
 are the only operations allowed on corrupt tables.
 .Pp
-If the first sector of a provider is corrupt, the kernel can not detect GPT
-even if the partition table itself is not corrupt.
-The protective MBR can be rewritten using the
-.Xr dd 1
-command, to restore the ability to detect the GPT.
-The copy of the protective MBR is usually located in the
-.Pa /boot/pmbr
-file.
-.Pp
 If one GPT header appears to be corrupt but the other copy remains intact,
 the kernel will log the following:
 .Bd -literal -offset indent
@@ -1330,7 +1321,6 @@ and
 /sbin/gpart backup ada0 | /sbin/gpart restore -F ada1 ada2
 .Ed
 .Sh SEE ALSO
-.Xr dd 1 ,
 .Xr geom 4 ,
 .Xr boot0cfg 8 ,
 .Xr geom 8 ,
diff --git a/sbin/ifconfig/ifconfig.c b/sbin/ifconfig/ifconfig.c
index bb9eb4a..1b89e67 100644
--- a/sbin/ifconfig/ifconfig.c
+++ b/sbin/ifconfig/ifconfig.c
@@ -855,7 +855,7 @@ setifmetric(const char *val, int dummy __unused, int s,
 	strncpy(ifr.ifr_name, name, sizeof (ifr.ifr_name));
 	ifr.ifr_metric = atoi(val);
 	if (ioctl(s, SIOCSIFMETRIC, (caddr_t)&ifr) < 0)
-		warn("ioctl (set metric)");
+		err(1, "ioctl SIOCSIFMETRIC (set metric)");
 }
 
 static void
@@ -865,7 +865,7 @@ setifmtu(const char *val, int dummy __unused, int s,
 	strncpy(ifr.ifr_name, name, sizeof (ifr.ifr_name));
 	ifr.ifr_mtu = atoi(val);
 	if (ioctl(s, SIOCSIFMTU, (caddr_t)&ifr) < 0)
-		warn("ioctl (set mtu)");
+		err(1, "ioctl SIOCSIFMTU (set mtu)");
 }
 
 static void
@@ -875,15 +875,12 @@ setifname(const char *val, int dummy __unused, int s,
 	char *newname;
 
 	newname = strdup(val);
-	if (newname == NULL) {
-		warn("no memory to set ifname");
-		return;
-	}
+	if (newname == NULL)
+		err(1, "no memory to set ifname");
 	ifr.ifr_data = newname;
 	if (ioctl(s, SIOCSIFNAME, (caddr_t)&ifr) < 0) {
-		warn("ioctl (set name)");
 		free(newname);
-		return;
+		err(1, "ioctl SIOCSIFNAME (set name)");
 	}
 	strlcpy(name, newname, sizeof(name));
 	free(newname);
@@ -910,7 +907,7 @@ setifdescr(const char *val, int dummy __unused, int s,
 	}
 
 	if (ioctl(s, SIOCSIFDESCR, (caddr_t)&ifr) < 0)
-		warn("ioctl (set descr)");
+		err(1, "ioctl SIOCSIFDESCR (set descr)");
 
 	free(newdescr);
 }
diff --git a/sbin/init/Makefile b/sbin/init/Makefile
index 7497a4b..eebaf59 100644
--- a/sbin/init/Makefile
+++ b/sbin/init/Makefile
@@ -2,6 +2,7 @@
 # $FreeBSD$
 
 PROG=	init
+SRCS=	init.c getmntopts.c
 MAN=	init.8
 PRECIOUSPROG=
 INSTALLFLAGS=-b -B.bak
@@ -9,6 +10,11 @@ CFLAGS+=-DDEBUGSHELL -DSECURE -DLOGIN_CAP -DCOMPAT_SYSV_INIT
 DPADD=	${LIBUTIL} ${LIBCRYPT}
 LDADD=	-lutil -lcrypt
 
+# Needed for getmntopts.c
+MOUNT=	${.CURDIR}/../../sbin/mount
+CFLAGS+=-I${MOUNT}
+.PATH:	${MOUNT}
+
 NO_SHARED?=	YES
 
 .include <bsd.prog.mk>
diff --git a/sbin/init/init.c b/sbin/init/init.c
index e6f567c..af672a9 100644
--- a/sbin/init/init.c
+++ b/sbin/init/init.c
@@ -46,6 +46,7 @@ static const char rcsid[] =
 
 #include <sys/param.h>
 #include <sys/ioctl.h>
+#include <sys/mman.h>
 #include <sys/mount.h>
 #include <sys/sysctl.h>
 #include <sys/wait.h>
@@ -79,6 +80,7 @@ static const char rcsid[] =
 #include <login_cap.h>
 #endif
 
+#include "mntopts.h"
 #include "pathnames.h"
 
 /*
@@ -103,6 +105,7 @@ static void warning(const char *, ...) __printflike(1, 2);
 static void emergency(const char *, ...) __printflike(1, 2);
 static void disaster(int);
 static void badsys(int);
+static void revoke_ttys(void);
 static int  runshutdown(void);
 static char *strk(char *);
 
@@ -122,6 +125,8 @@ static state_func_t clean_ttys(void);
 static state_func_t catatonia(void);
 static state_func_t death(void);
 static state_func_t death_single(void);
+static state_func_t reroot(void);
+static state_func_t reroot_phase_two(void);
 
 static state_func_t run_script(const char *);
 
@@ -194,7 +199,7 @@ main(int argc, char *argv[])
 {
 	state_t initial_transition = runcom;
 	char kenv_value[PATH_MAX];
-	int c;
+	int c, error;
 	struct sigaction sa;
 	sigset_t mask;
 
@@ -227,6 +232,9 @@ main(int argc, char *argv[])
 				case 'q': /* rescan /etc/ttys */
 					sig = SIGHUP;
 					break;
+				case 'r': /* remount root */
+					sig = SIGEMT;
+					break;
 				default:
 					goto invalid;
 				}
@@ -248,7 +256,7 @@ invalid:
 	/*
 	 * Create an initial session.
 	 */
-	if (setsid() < 0)
+	if (setsid() < 0 && (errno != EPERM || getsid(0) != 1))
 		warning("initial setsid() failed: %m");
 
 	/*
@@ -262,7 +270,7 @@ invalid:
 	 * This code assumes that we always get arguments through flags,
 	 * never through bits set in some random machine register.
 	 */
-	while ((c = getopt(argc, argv, "dsf")) != -1)
+	while ((c = getopt(argc, argv, "dsfr")) != -1)
 		switch (c) {
 		case 'd':
 			devfs = 1;
@@ -273,6 +281,9 @@ invalid:
 		case 'f':
 			runcom_mode = FASTBOOT;
 			break;
+		case 'r':
+			initial_transition = reroot_phase_two;
+			break;
 		default:
 			warning("unrecognized flag '-%c'", c);
 			break;
@@ -288,13 +299,13 @@ invalid:
 	handle(badsys, SIGSYS, 0);
 	handle(disaster, SIGABRT, SIGFPE, SIGILL, SIGSEGV, SIGBUS, SIGXCPU,
 	    SIGXFSZ, 0);
-	handle(transition_handler, SIGHUP, SIGINT, SIGTERM, SIGTSTP, SIGUSR1,
-	    SIGUSR2, 0);
+	handle(transition_handler, SIGHUP, SIGINT, SIGEMT, SIGTERM, SIGTSTP,
+	    SIGUSR1, SIGUSR2, 0);
 	handle(alrm_handler, SIGALRM, 0);
 	sigfillset(&mask);
 	delset(&mask, SIGABRT, SIGFPE, SIGILL, SIGSEGV, SIGBUS, SIGSYS,
-	    SIGXCPU, SIGXFSZ, SIGHUP, SIGINT, SIGTERM, SIGTSTP, SIGALRM,
-	    SIGUSR1, SIGUSR2, 0);
+	    SIGXCPU, SIGXFSZ, SIGHUP, SIGINT, SIGEMT, SIGTERM, SIGTSTP,
+	    SIGALRM, SIGUSR1, SIGUSR2, 0);
 	sigprocmask(SIG_SETMASK, &mask, (sigset_t *) 0);
 	sigemptyset(&sa.sa_mask);
 	sa.sa_flags = 0;
@@ -374,6 +385,16 @@ invalid:
 			free(s);
 	}
 
+	if (initial_transition != reroot_phase_two) {
+		/*
+		 * Unmount reroot leftovers.  This runs after init(8)
+		 * gets reexecuted after reroot_phase_two() is done.
+		 */
+		error = unmount(_PATH_REROOT, MNT_FORCE);
+		if (error != 0 && errno != EINVAL)
+			warning("Cannot unmount %s: %m", _PATH_REROOT);
+	}
+
 	/*
 	 * Start the state machine.
 	 */
@@ -621,6 +642,229 @@ write_stderr(const char *message)
 	write(STDERR_FILENO, message, strlen(message));
 }
 
+static int
+read_file(const char *path, void **bufp, size_t *bufsizep)
+{
+	struct stat sb;
+	size_t bufsize;
+	void *buf;
+	ssize_t nbytes;
+	int error, fd;
+
+	fd = open(path, O_RDONLY);
+	if (fd < 0) {
+		emergency("%s: %s", path, strerror(errno));
+		return (-1);
+	}
+
+	error = fstat(fd, &sb);
+	if (error != 0) {
+		emergency("fstat: %s", strerror(errno));
+		close(fd);
+		return (error);
+	}
+
+	bufsize = sb.st_size;
+	buf = malloc(bufsize);
+	if (buf == NULL) {
+		emergency("malloc: %s", strerror(errno));
+		close(fd);
+		return (error);
+	}
+
+	nbytes = read(fd, buf, bufsize);
+	if (nbytes != (ssize_t)bufsize) {
+		emergency("read: %s", strerror(errno));
+		close(fd);
+		free(buf);
+		return (error);
+	}
+
+	error = close(fd);
+	if (error != 0) {
+		emergency("close: %s", strerror(errno));
+		free(buf);
+		return (error);
+	}
+
+	*bufp = buf;
+	*bufsizep = bufsize;
+
+	return (0);
+}
+
+static int
+create_file(const char *path, const void *buf, size_t bufsize)
+{
+	ssize_t nbytes;
+	int error, fd;
+
+	fd = open(path, O_WRONLY | O_CREAT | O_EXCL, 0700);
+	if (fd < 0) {
+		emergency("%s: %s", path, strerror(errno));
+		return (-1);
+	}
+
+	nbytes = write(fd, buf, bufsize);
+	if (nbytes != (ssize_t)bufsize) {
+		emergency("write: %s", strerror(errno));
+		close(fd);
+		return (-1);
+	}
+
+	error = close(fd);
+	if (error != 0) {
+		emergency("close: %s", strerror(errno));
+		return (-1);
+	}
+
+	return (0);
+}
+
+static int
+mount_tmpfs(const char *fspath)
+{
+	struct iovec *iov;
+	char errmsg[255];
+	int error, iovlen;
+
+	iov = NULL;
+	iovlen = 0;
+	memset(errmsg, 0, sizeof(errmsg));
+	build_iovec(&iov, &iovlen, "fstype",
+	    __DECONST(void *, "tmpfs"), (size_t)-1);
+	build_iovec(&iov, &iovlen, "fspath",
+	    __DECONST(void *, fspath), (size_t)-1);
+	build_iovec(&iov, &iovlen, "errmsg",
+	    errmsg, sizeof(errmsg));
+
+	error = nmount(iov, iovlen, 0);
+	if (error != 0) {
+		if (*errmsg != '\0') {
+			emergency("cannot mount tmpfs on %s: %s: %s",
+			    fspath, errmsg, strerror(errno));
+		} else {
+			emergency("cannot mount tmpfs on %s: %s",
+			    fspath, strerror(errno));
+		}
+		return (error);
+	}
+	return (0);
+}
+
+static state_func_t
+reroot(void)
+{
+	void *buf;
+	char init_path[PATH_MAX];
+	size_t bufsize, init_path_len;
+	int error, name[4];
+
+	buf = NULL;
+	bufsize = 0;
+
+	name[0] = CTL_KERN;
+	name[1] = KERN_PROC;
+	name[2] = KERN_PROC_PATHNAME;
+	name[3] = -1;
+	init_path_len = sizeof(init_path);
+	error = sysctl(name, 4, init_path, &init_path_len, NULL, 0);
+	if (error != 0) {
+		emergency("failed to get kern.proc.pathname: %s",
+		    strerror(errno));
+		goto out;
+	}
+
+	revoke_ttys();
+	runshutdown();
+
+	/*
+	 * Make sure nobody can interfere with our scheme.
+	 */
+	error = kill(-1, SIGKILL);
+	if (error != 0) {
+		emergency("kill(2) failed: %s", strerror(errno));
+		goto out;
+	}
+
+	/*
+	 * Copy the init binary into tmpfs, so that we can unmount
+	 * the old rootfs without committing suicide.
+	 */
+	error = read_file(init_path, &buf, &bufsize);
+	if (error != 0)
+		goto out;
+	error = mount_tmpfs(_PATH_REROOT);
+	if (error != 0)
+		goto out;
+	error = create_file(_PATH_REROOT_INIT, buf, bufsize);
+	if (error != 0)
+		goto out;
+
+	/*
+	 * Execute the temporary init.
+	 */
+	execl(_PATH_REROOT_INIT, _PATH_REROOT_INIT, "-r", NULL);
+	emergency("cannot exec %s: %s", _PATH_REROOT_INIT, strerror(errno));
+
+out:
+	emergency("reroot failed; going to single user mode");
+	free(buf);
+	return (state_func_t) single_user;
+}
+
+static state_func_t
+reroot_phase_two(void)
+{
+	char init_path[PATH_MAX], *path, *path_component;
+	size_t init_path_len;
+	int nbytes, error;
+
+	/*
+	 * Ask the kernel to mount the new rootfs.
+	 */
+	error = reboot(RB_REROOT);
+	if (error != 0) {
+		emergency("RB_REBOOT failed: %s", strerror(errno));
+		goto out;
+	}
+
+	/*
+	 * Figure out where the destination init(8) binary is.  Note that
+	 * the path could be different than what we've started with.  Use
+	 * the value from kenv, if set, or the one from sysctl otherwise.
+	 * The latter defaults to a hardcoded value, but can be overridden
+	 * by a build time option.
+	 */
+	nbytes = kenv(KENV_GET, "init_path", init_path, sizeof(init_path));
+	if (nbytes <= 0) {
+		init_path_len = sizeof(init_path);
+		error = sysctlbyname("kern.init_path",
+		    init_path, &init_path_len, NULL, 0);
+		if (error != 0) {
+			emergency("failed to retrieve kern.init_path: %s",
+			    strerror(errno));
+			goto out;
+		}
+	}
+
+	/*
+	 * Repeat the init search logic from sys/kern/init_path.c
+	 */
+	path_component = init_path;
+	while ((path = strsep(&path_component, ":")) != NULL) {
+		/*
+		 * Execute init(8) from the new rootfs.
+		 */
+		execl(path, path, NULL);
+	}
+	emergency("cannot exec init from %s: %s", init_path, strerror(errno));
+
+out:
+	emergency("reroot failed; going to single user mode");
+	return (state_func_t) single_user;
+}
+
 /*
  * Bring the system up single user.
  */
@@ -852,8 +1096,9 @@ run_script(const char *script)
 		if ((wpid = waitpid(-1, &status, WUNTRACED)) != -1)
 			collect_child(wpid);
 		if (wpid == -1) {
-			if (requested_transition == death_single)
-				return (state_func_t) death_single;
+			if (requested_transition == death_single ||
+			    requested_transition == reroot)
+				return (state_func_t) requested_transition;
 			if (errno == EINTR)
 				continue;
 			warning("wait for %s on %s failed: %m; going to "
@@ -1326,6 +1571,9 @@ transition_handler(int sig)
 		    current_state == multi_user || current_state == catatonia)
 			requested_transition = catatonia;
 		break;
+	case SIGEMT:
+		requested_transition = reroot;
+		break;
 	default:
 		requested_transition = 0;
 		break;
@@ -1498,7 +1746,6 @@ alrm_handler(int sig)
 static state_func_t
 death(void)
 {
-	session_t *sp;
 	int block, blocked;
 	size_t len;
 
@@ -1515,11 +1762,7 @@ death(void)
 	 * runshutdown() will perform the initial open() call, causing
 	 * the terminal attributes to be misconfigured.
 	 */
-	for (sp = sessions; sp; sp = sp->se_next) {
-		sp->se_flags |= SE_SHUTDOWN;
-		kill(sp->se_process, SIGHUP);
-		revoke(sp->se_device);
-	}
+	revoke_ttys();
 
 	/* Try to run the rc.shutdown script within a period of time */
 	runshutdown();
@@ -1565,6 +1808,18 @@ death_single(void)
 	return (state_func_t) single_user;
 }
 
+static void
+revoke_ttys(void)
+{
+	session_t *sp;
+
+	for (sp = sessions; sp; sp = sp->se_next) {
+		sp->se_flags |= SE_SHUTDOWN;
+		kill(sp->se_process, SIGHUP);
+		revoke(sp->se_device);
+	}
+}
+
 /*
  * Run the system shutdown script.
  *
diff --git a/sbin/init/pathnames.h b/sbin/init/pathnames.h
index 39eed4c..ec33b45 100644
--- a/sbin/init/pathnames.h
+++ b/sbin/init/pathnames.h
@@ -35,7 +35,9 @@
 
 #include <paths.h>
 
-#define	_PATH_INITLOG	"/var/log/init.log"
-#define	_PATH_SLOGGER	"/sbin/session_logger"
-#define	_PATH_RUNCOM	"/etc/rc"
-#define _PATH_RUNDOWN   "/etc/rc.shutdown"
+#define	_PATH_INITLOG		"/var/log/init.log"
+#define	_PATH_SLOGGER		"/sbin/session_logger"
+#define	_PATH_RUNCOM		"/etc/rc"
+#define	_PATH_RUNDOWN		"/etc/rc.shutdown"
+#define	_PATH_REROOT		"/dev/reroot"
+#define	_PATH_REROOT_INIT	_PATH_REROOT "/init"
diff --git a/sbin/mdconfig/mdconfig.8 b/sbin/mdconfig/mdconfig.8
index 8cb9a69..d273128 100644
--- a/sbin/mdconfig/mdconfig.8
+++ b/sbin/mdconfig/mdconfig.8
@@ -41,7 +41,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd August 6, 2015
+.Dd October 10, 2015
 .Dt MDCONFIG 8
 .Os
 .Sh NAME
@@ -163,11 +163,11 @@ prefix.
 Size of the memory disk.
 .Ar Size
 is the number of 512 byte sectors unless suffixed with a
-.Cm b , k , m , g ,
+.Cm b , k , m , g , t ,
 or
-.Cm t
+.Cm p
 which
-denotes byte, kilobyte, megabyte, gigabyte and terabyte respectively.
+denotes byte, kilobyte, megabyte, gigabyte, terabyte and petabyte respectively.
 When used without the
 .Fl r
 option, the
@@ -304,8 +304,7 @@ mount /dev/md1.nop /mnt
 .Sh SEE ALSO
 .Xr md 4 ,
 .Xr ffs 7 ,
-.Xr bsdlabel 8 ,
-.Xr fdisk 8 ,
+.Xr gpart 8 ,
 .Xr mdmfs 8 ,
 .Xr malloc 9
 .Sh HISTORY
diff --git a/sbin/mdconfig/mdconfig.c b/sbin/mdconfig/mdconfig.c
index d741c77..1b4e4ce 100644
--- a/sbin/mdconfig/mdconfig.c
+++ b/sbin/mdconfig/mdconfig.c
@@ -88,8 +88,8 @@ usage(void)
 	fprintf(stderr, "\t\ttype = {malloc, vnode, swap}\n");
 	fprintf(stderr, "\t\toption = {cluster, compress, reserve}\n");
 	fprintf(stderr, "\t\tsize = %%d (512 byte blocks), %%db (B),\n");
-	fprintf(stderr, "\t\t       %%dk (kB), %%dm (MB), %%dg (GB) or\n");
-	fprintf(stderr, "\t\t       %%dt (TB)\n");
+	fprintf(stderr, "\t\t       %%dk (kB), %%dm (MB), %%dg (GB), \n");
+	fprintf(stderr, "\t\t       %%dt (TB), or %%dp (PB)\n");
 	exit(1);
 }
 
@@ -217,6 +217,9 @@ main(int argc, char **argv)
 			else if (*p == 't' || *p == 'T') {
 				mdio.md_mediasize <<= 30;
 				mdio.md_mediasize <<= 10;
+			} else if (*p == 'p' || *p == 'P') {
+				mdio.md_mediasize <<= 30;
+				mdio.md_mediasize <<= 20;
 			} else
 				errx(1, "unknown suffix on -s argument");
 			break;
diff --git a/sbin/newfs/newfs.8 b/sbin/newfs/newfs.8
index 6764adc..f1bdd6b 100644
--- a/sbin/newfs/newfs.8
+++ b/sbin/newfs/newfs.8
@@ -303,11 +303,11 @@ on file systems that contain many small files.
 .Xr geom 4 ,
 .Xr disktab 5 ,
 .Xr fs 5 ,
-.Xr bsdlabel 8 ,
 .Xr camcontrol 8 ,
 .Xr dump 8 ,
 .Xr dumpfs 8 ,
 .Xr fsck 8 ,
+.Xr gpart 8 ,
 .Xr gjournal 8 ,
 .Xr growfs 8 ,
 .Xr makefs 8 ,
diff --git a/sbin/newfs_msdos/newfs_msdos.8 b/sbin/newfs_msdos/newfs_msdos.8
index 218da93..8901673 100644
--- a/sbin/newfs_msdos/newfs_msdos.8
+++ b/sbin/newfs_msdos/newfs_msdos.8
@@ -228,9 +228,7 @@ Create a 30MB image file, with the FAT partition starting
 newfs_msdos -C 30M -@63s ./somefile
 .Ed
 .Sh SEE ALSO
-.Xr disktab 5 ,
-.Xr disklabel 8 ,
-.Xr fdisk 8 ,
+.Xr gpart 8 ,
 .Xr newfs 8
 .Sh HISTORY
 The
diff --git a/sbin/newfs_nandfs/newfs_nandfs.8 b/sbin/newfs_nandfs/newfs_nandfs.8
index 6997430..b357547 100644
--- a/sbin/newfs_nandfs/newfs_nandfs.8
+++ b/sbin/newfs_nandfs/newfs_nandfs.8
@@ -63,9 +63,7 @@ Create a file system, using default parameters, on
 newfs_nandfs /dev/ada0s1
 .Ed
 .Sh SEE ALSO
-.Xr disktab 5 ,
-.Xr disklabel 8 ,
-.Xr fdisk 8 ,
+.Xr gpart 8 ,
 .Xr newfs 8
 .Sh HISTORY
 The
diff --git a/sbin/reboot/boot_i386.8 b/sbin/reboot/boot_i386.8
index e21e53f..b86e318 100644
--- a/sbin/reboot/boot_i386.8
+++ b/sbin/reboot/boot_i386.8
@@ -348,9 +348,9 @@ requirement has not been adhered to.
 .Xr make.conf 5 ,
 .Xr ttys 5 ,
 .Xr boot0cfg 8 ,
-.Xr bsdlabel 8 ,
 .Xr btxld 8 ,
 .Xr config 8 ,
+.Xr gpart 8 ,
 .Xr gptboot 8 ,
 .Xr halt 8 ,
 .Xr loader 8 ,
diff --git a/sbin/reboot/reboot.8 b/sbin/reboot/reboot.8
index 84dd887..420d4f8 100644
--- a/sbin/reboot/reboot.8
+++ b/sbin/reboot/reboot.8
@@ -28,7 +28,7 @@
 .\"	@(#)reboot.8	8.1 (Berkeley) 6/9/93
 .\" $FreeBSD$
 .\"
-.Dd October 11, 2010
+.Dd May 22, 2015
 .Dd Jan 06, 2016
 .Dt REBOOT 8
 .Os
@@ -43,7 +43,7 @@
 .Op Fl lNnpq
 .Op Fl k Ar kernel
 .Nm
-.Op Fl dlNnpq
+.Op Fl dlNnpqr
 .Op Fl k Ar kernel
 .Nm fasthalt
 .Op Fl lNnpq
@@ -122,6 +122,13 @@ the flushing of the file system cache is performed (if the
 .Fl n
 option is not specified).
 This option should probably not be used.
+.It Fl r
+The system kills all processes, unmounts all filesystems, mounts the new
+root filesystem, and begins the usual startup sequence.
+After changing vfs.root.mountfrom with
+.Xr kenv 8 ,
+.Nm Fl r
+can be used to change the root filesystem while preserving kernel state.
 .El
 .Pp
 The
@@ -139,6 +146,13 @@ Normally, the
 utility is used when the system needs to be halted or restarted, giving
 users advance warning of their impending doom and cleanly terminating
 specific programs.
+.Sh EXAMPLES
+Replace current root filesystem with UFS mounted from
+.Pa /dev/ada0s1a :
+.Bd -literal -offset indent
+kenv vfs.root.mountfrom=ufs:/dev/ada0s1a
+reboot -r
+.Ed
 .Sh SEE ALSO
 .Xr getutxent 3 ,
 .Xr boot 8 ,
diff --git a/sbin/reboot/reboot.c b/sbin/reboot/reboot.c
index 6d590ac..961ba41 100644
--- a/sbin/reboot/reboot.c
+++ b/sbin/reboot/reboot.c
@@ -77,7 +77,7 @@ main(int argc, char *argv[])
 	} else
 		howto = 0;
 	lflag = nflag = qflag = Nflag = 0;
-	while ((ch = getopt(argc, argv, "dk:lNnpq")) != -1)
+	while ((ch = getopt(argc, argv, "dk:lNnpqr")) != -1)
 		switch(ch) {
 		case 'd':
 			howto |= RB_DUMP;
@@ -102,6 +102,9 @@ main(int argc, char *argv[])
 		case 'q':
 			qflag = 1;
 			break;
+		case 'r':
+			howto |= RB_REROOT;
+			break;
 		case '?':
 		default:
 			usage();
@@ -113,6 +116,8 @@ main(int argc, char *argv[])
 		errx(1, "cannot dump (-d) when halting; must reboot instead");
 	if (Nflag && (howto & RB_NOSYNC) != 0)
 		errx(1, "-N cannot be used with -n");
+	if ((howto & RB_REROOT) != 0 && howto != RB_REROOT)
+		errx(1, "-r cannot be used with -d, -n, or -p");
 	if (geteuid()) {
 		errno = EPERM;
 		err(1, NULL);
@@ -143,6 +148,9 @@ main(int argc, char *argv[])
 		if (dohalt) {
 			openlog("halt", 0, LOG_AUTH | LOG_CONS);
 			syslog(LOG_CRIT, "halted by %s", user);
+		} else if (howto & RB_REROOT) {
+			openlog("reroot", 0, LOG_AUTH | LOG_CONS);
+			syslog(LOG_CRIT, "rerooted by %s", user);
 		} else {
 			openlog("reboot", 0, LOG_AUTH | LOG_CONS);
 			syslog(LOG_CRIT, "rebooted by %s", user);
@@ -176,6 +184,16 @@ main(int argc, char *argv[])
 	 */
 	(void)signal(SIGPIPE, SIG_IGN);
 
+	/*
+	 * Only init(8) can perform rerooting.
+	 */
+	if (howto & RB_REROOT) {
+		if (kill(1, SIGEMT) == -1)
+			err(1, "SIGEMT init");
+
+		return (0);
+	}
+
 	/* Just stop init -- if we fail, we'll restart it. */
 	if (kill(1, SIGTSTP) == -1)
 		err(1, "SIGTSTP init");