unshare: add --propagation, use MS_PRIVATE by default

After "unshare --mount" users assume that mount operations within the
new namespaces are unshared (invisible for the rest of the system).

Unfortunately, this is not true and the behavior depends on the
current mount propagation setting. The kernel default is "private",
but for example systemd based distros use "shared". The solution is to
use (for example) "mount --make-private" after unshare(1).

I have been requested many times to provide less fragile and more
unified unshared mount setting *by default* to make things user
friendly.

The patch forces unshare(1) to explicitly use MS_REC|MS_PRIVATE for all
tree by default.

We can use something less (e.g MS_SLAVE), but "private" is the kernel
default, so for many users this change (feature) will be invisible.

This feature is possible to disable by "--propagation unchanged" or it's
possible to specify another propagation flag, supported are:

	<slave|shared|private|unchanged>

Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

sys-utils/unshare.1

@@ -14,12 +14,14 @@ options.  Unshareable namespaces are:
 .BR "mount namespace"
 Mounting and unmounting filesystems will not affect the rest of the system
 (\fBCLONE_NEWNS\fP flag), except for filesystems which are explicitly marked as
-shared (with \fBmount --make-shared\fP; see \fI/proc/self/mountinfo\fP for the
-\fBshared\fP flags).
+shared (with \fBmount --make-shared\fP; see \fI/proc/self/mountinfo\fP or
+\fBfindmnt -o+PROPAGATION\fP for the \fBshared\fP flags).
 .sp
-It's recommended to use \fBmount --make-rprivate\fP or \fBmount --make-rslave\fP
-after \fBunshare --mount\fP to make sure that mountpoints in the new namespace
-are really unshared from the parental namespace.
+.B unshare
+since util-linux version 2.27 automatically sets propagation to \fBprivate\fP
+in the new mount namespace to make sure that the new namespace is really
+unshared. This feature is possible to disable by option \fB\-\-propagation unchanged\fP.
+Note that \fBprivate\fP is the kernel default.
 .TP
 .BR "UTS namespace"
 Setting hostname or domainname will not affect the rest of the system.
@@ -84,7 +86,13 @@ the mount namespace) even when run unprivileged.  As a mere convenience feature,
 more sophisticated use cases, such as mapping multiple ranges of UIDs and GIDs.
 This option implies --setgroups=deny.
 .TP
-.BR \-s , " \-\-setgroups \fIallow|deny\fP"
+.BR "\-\-propagation \fIprivate|shared|slave|unchanged\fP"
+Recursively sets mount propagation flag in the new mount namespace. The default
+is to set the propagation to \fIprivate\fP, this feature is possible to disable
+by \fIunchanged\fP argument. The options is silently ignored when mount namespace (\fB\-\-mount\fP)
+is not requested.
+.TP
+.BR "\-\-setgroups \fIallow|deny\fP"
 Allow or deny
 .BR setgroups (2)
 syscall in user namespaces.

sys-utils/unshare.c

@@ -39,6 +39,9 @@
 #include "pathnames.h"
 #include "all-io.h"
 
+/* 'private' is kernel default */
+#define UNSHARE_PROPAGATION_DEFAULT	(MS_REC | MS_PRIVATE)
+
 enum {
 	SETGROUPS_NONE = -1,
 	SETGROUPS_DENY = 0,
@@ -100,6 +103,36 @@ static void map_id(const char *file, uint32_t from, uint32_t to)
 	close(fd);
 }
 
+static unsigned long parse_propagation(const char *str)
+{
+	size_t i;
+	static const struct prop_opts {
+		const char *name;
+		unsigned long flag;
+	} opts[] = {
+		{ "slave",	MS_REC | MS_SLAVE },
+		{ "private",	MS_REC | MS_PRIVATE },
+		{ "shared",     MS_REC | MS_SHARED },
+		{ "unchanged",        0 }
+	};
+
+	for (i = 0; i < ARRAY_SIZE(opts); i++) {
+		if (strcmp(opts[i].name, str) == 0)
+			return opts[i].flag;
+	}
+
+	errx(EXIT_FAILURE, _("unsupported propagation mode: %s"), str);
+}
+
+static void set_propagation(unsigned long flags)
+{
+	if (flags == 0)
+		return;
+
+	if (mount("none", "/", NULL, flags, NULL) != 0)
+		err(EXIT_FAILURE, _("cannot change root filesystem propagation"));
+}
+
 static void usage(int status)
 {
 	FILE *out = status == EXIT_SUCCESS ? stdout : stderr;
@@ -121,6 +154,8 @@ static void usage(int status)
 	fputs(_(" -f, --fork                fork before launching <program>\n"), out);
 	fputs(_("     --mount-proc[=<dir>]  mount proc filesystem first (implies --mount)\n"), out);
 	fputs(_(" -r, --map-root-user       map current user to root (implies --user)\n"), out);
+	fputs(_("     --propagation <slave|shared|private|unchanged>\n"
+	        "                           modify mount propagation in mount namespace\n"), out);
 	fputs(_(" -s, --setgroups allow|deny  control the setgroups syscall in user namespaces\n"), out);
 
 	fputs(USAGE_SEPARATOR, out);
@@ -135,6 +170,7 @@ int main(int argc, char *argv[])
 {
 	enum {
 		OPT_MOUNTPROC = CHAR_MAX + 1,
+		OPT_PROPAGATION,
 		OPT_SETGROUPS
 	};
 	static const struct option longopts[] = {
@@ -149,6 +185,7 @@ int main(int argc, char *argv[])
 		{ "fork", no_argument, 0, 'f' },
 		{ "mount-proc", optional_argument, 0, OPT_MOUNTPROC },
 		{ "map-root-user", no_argument, 0, 'r' },
+		{ "propagation", required_argument, 0, OPT_PROPAGATION },
 		{ "setgroups", required_argument, 0, OPT_SETGROUPS },
 		{ NULL, 0, 0, 0 }
 	};
@@ -157,6 +194,7 @@ int main(int argc, char *argv[])
 	int unshare_flags = 0;
 	int c, forkit = 0, maproot = 0;
 	const char *procmnt = NULL;
+	unsigned long propagation = UNSHARE_PROPAGATION_DEFAULT;
 	uid_t real_euid = geteuid();
 	gid_t real_egid = getegid();;
 
@@ -204,6 +242,9 @@ int main(int argc, char *argv[])
 		case OPT_SETGROUPS:
 			setgrpcmd = setgroups_str2id(optarg);
 			break;
+		case OPT_PROPAGATION:
+			propagation = parse_propagation(optarg);
+			break;
 		default:
 			usage(EXIT_FAILURE);
 		}
@@ -248,6 +289,9 @@ int main(int argc, char *argv[])
 	} else if (setgrpcmd != SETGROUPS_NONE)
 		setgroups_control(setgrpcmd);
 
+	if ((unshare_flags & CLONE_NEWNS) && propagation)
+		set_propagation(propagation);
+
 	if (procmnt &&
 	    (mount("none", procmnt, NULL, MS_PRIVATE|MS_REC, NULL) != 0 ||
 	     mount("proc", procmnt, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL) != 0))