Util-Linux Archive on lore.kernel.org
 help / color / Atom feed
From: Karel Zak <kzak@redhat.com>
To: util-linux@vger.kernel.org
Subject: [HEADS-UP]: mount(8) non-root user mount changes
Date: Tue, 19 Nov 2019 15:34:08 +0100
Message-ID: <20191119143408.b6pcbhtsiubxdcob@10.255.255.10> (raw)


 Hi,

 for years users are unhappy with too strict mount/umount rules for
 non-root users.  The rules makes mount(8) useless for use-cases where
 root permissions are unnecessary. This patch changes the game.
 
 Please, review, complain, ...

    Karel

From 6497f2d99e9cabee3531e644ba4dcffd14532200 Mon Sep 17 00:00:00 2001
From: Karel Zak <kzak@redhat.com>
Date: Tue, 19 Nov 2019 14:58:20 +0100
Subject: [PATCH] mount: no exit on EPERM, continue without suid
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The current libmount assumes that mount(8) and umount(8) are suid
binaries. For this reason it implements internal rules which
restrict what is allowed for non-root users. Unfortunately, it's
out of reality for some use-cases where root permissions are no
required. Nice example are fuse filesystems.

So, the current situation is to call exit() always when mount, umount or
libmount are unsure with non-root user rights. This patch removes the
exit() call and replaces it with suid permissions drop, after that it
continues as usually. It means after suid-drop all depend on kernel
and no another security rule is used by libmount (simply because any
rule is no more necessary).

Example:

old version:
   $ mount -t fuse.sshfs kzak@192.168.111.1:/home/kzak /home/kzak/mnt
   mount: only root can use "--types" option

new version:
   $ mount -t fuse.sshfs kzak@192.168.111.1:/home/kzak /home/kzak/mnt
   kzak@192.168.111.1's password:

   $ findmnt /home/kzak/mnt
   TARGET         SOURCE                        FSTYPE     OPTIONS
   /home/kzak/mnt kzak@192.168.111.1:/home/kzak fuse.sshfs rw,nosuid,nodev,relatime,user_id=1000,group_id=1000

   $ umount /home/kzak/mnt
   $ echo $?
   0

Note that fuse user umount is supported since v2.34 due to user_id= in
kernel mount table.

Signed-off-by: Karel Zak <kzak@redhat.com>
---
 libmount/docs/libmount-sections.txt |  1 +
 libmount/src/context.c              | 25 +++++++++++++++++
 libmount/src/libmount.h.in          |  1 +
 libmount/src/libmount.sym           |  1 +
 sys-utils/mount.8                   |  6 ++++
 sys-utils/mount.c                   | 43 +++++++++++++++++------------
 sys-utils/umount.8                  | 19 +++++++++++++
 sys-utils/umount.c                  | 37 +++++++++++++++----------
 8 files changed, 102 insertions(+), 31 deletions(-)

diff --git a/libmount/docs/libmount-sections.txt b/libmount/docs/libmount-sections.txt
index 990c0394f..0bba9f835 100644
--- a/libmount/docs/libmount-sections.txt
+++ b/libmount/docs/libmount-sections.txt
@@ -40,6 +40,7 @@ mnt_context_enable_rwonly_mount
 mnt_context_enable_sloppy
 mnt_context_enable_verbose
 mnt_context_forced_rdonly
+mnt_context_force_unrestricted
 mnt_context_get_cache
 mnt_context_get_excode
 mnt_context_get_fs
diff --git a/libmount/src/context.c b/libmount/src/context.c
index 977842f7d..72a0c5a7b 100644
--- a/libmount/src/context.c
+++ b/libmount/src/context.c
@@ -426,6 +426,31 @@ int mnt_context_is_restricted(struct libmnt_context *cxt)
 	return cxt->restricted;
 }
 
+/**
+ * mnt_context_force_unrestricted:
+ * @cxt: mount context
+ *
+ * This function is DANGEROURS as it disables all security policies in libmount.
+ * Don't use if not sure. It removes "restricted" flag from the context, so
+ * libmount will use the current context as for root user.
+ *
+ * This function is designed for case you have no any suid permissions, so you
+ * can depend on kernel.
+ *
+ * Returns: 0 on success, negative number in case of error.
+ *
+ * Since: 2.35
+ */
+int mnt_context_force_unrestricted(struct libmnt_context *cxt)
+{
+	if (mnt_context_is_restricted(cxt)) {
+		DBG(CXT, ul_debugobj(cxt, "force UNRESTRICTED"));
+		cxt->restricted = 0;
+	}
+
+	return 0;
+}
+
 /**
  * mnt_context_set_optsmode
  * @cxt: mount context
diff --git a/libmount/src/libmount.h.in b/libmount/src/libmount.h.in
index 19d4c5b53..ba54cf25d 100644
--- a/libmount/src/libmount.h.in
+++ b/libmount/src/libmount.h.in
@@ -705,6 +705,7 @@ extern void mnt_free_context(struct libmnt_context *cxt);
 extern int mnt_reset_context(struct libmnt_context *cxt);
 extern int mnt_context_is_restricted(struct libmnt_context *cxt)
 			__ul_attribute__((nonnull));
+extern int mnt_context_force_unrestricted(struct libmnt_context *cxt);
 
 extern int mnt_context_init_helper(struct libmnt_context *cxt,
 				   int action, int flags);
diff --git a/libmount/src/libmount.sym b/libmount/src/libmount.sym
index b9a4c0d22..792d11753 100644
--- a/libmount/src/libmount.sym
+++ b/libmount/src/libmount.sym
@@ -352,6 +352,7 @@ MOUNT_2.34 {
 } MOUNT_2.33;
 
 MOUNT_2_35 {
+	mnt_context_force_unrestricted;
 	mnt_context_get_target_prefix;
 	mnt_context_set_target_prefix;
 } MOUNT_2.34;
diff --git a/sys-utils/mount.8 b/sys-utils/mount.8
index 9d31d8245..a6231c7c1 100644
--- a/sys-utils/mount.8
+++ b/sys-utils/mount.8
@@ -315,6 +315,12 @@ program is executed. It's strongly recommended to use a valid mountpoint to
 specify filesystem, otherwise \fBmount\fR may fail. For example it's bad idea
 to use NFS or CIFS source on command line.
 .PP
+Since version 2.35 \fBmount\fR command does not exit when user permissions are
+inadequate by internal libmount security rules.  It drops suid permissions
+and continue as regular non-root user. It allows to support use-cases where
+root permissions are not necessary (e.g. fuse filesystems, user namespaces,
+etc).
+.PP
 For more details, see
 .BR fstab (5).
 Only the user that mounted a filesystem can unmount it again.
diff --git a/sys-utils/mount.c b/sys-utils/mount.c
index 08da9e6a5..5842bc2ec 100644
--- a/sys-utils/mount.c
+++ b/sys-utils/mount.c
@@ -47,23 +47,24 @@
 
 static int mk_exit_code(struct libmnt_context *cxt, int rc);
 
-static void __attribute__((__noreturn__)) exit_non_root(const char *option)
+static void suid_drop(struct libmnt_context *cxt)
 {
 	const uid_t ruid = getuid();
 	const uid_t euid = geteuid();
 
-	if (ruid == 0 && euid != 0) {
-		/* user is root, but setuid to non-root */
-		if (option)
-			errx(MNT_EX_USAGE, _("only root can use \"--%s\" option "
-					 "(effective UID is %u)"),
-					option, euid);
-		errx(MNT_EX_USAGE, _("only root can do that "
-				 "(effective UID is %u)"), euid);
+	if (ruid != 0 && euid == 0) {
+		if (setgid(getgid()) < 0)
+			err(MNT_EX_FAIL, _("setgid() failed"));
+
+		if (setuid(getuid()) < 0)
+			err(MNT_EX_FAIL, _("setuid() failed"));
 	}
-	if (option)
-		errx(MNT_EX_USAGE, _("only root can use \"--%s\" option"), option);
-	errx(MNT_EX_USAGE, _("only root can do that"));
+
+	/* be paranoid and check it, setuid(0) has to fail */
+	if (ruid != 0 && setuid(0) == 0)
+		errx(MNT_EX_FAIL, _("drop permissions failed."));
+
+	mnt_context_force_unrestricted(cxt);
 }
 
 static void __attribute__((__noreturn__)) mount_print_version(void)
@@ -672,7 +673,7 @@ int main(int argc, char **argv)
 		    !strchr("hlLUVvrist", c) &&
 		    c != MOUNT_OPT_TARGET &&
 		    c != MOUNT_OPT_SOURCE)
-			exit_non_root(option_to_longopt(c, longopts));
+			suid_drop(cxt);
 
 		err_exclusive_options(c, longopts, excl, excl_st);
 
@@ -872,7 +873,7 @@ int main(int argc, char **argv)
 	/* Non-root users are allowed to use -t to print_all(),
 	   but not to mount */
 	if (mnt_context_is_restricted(cxt) && types)
-		exit_non_root("types");
+		suid_drop(cxt);
 
 	if (oper && (types || all || mnt_context_get_source(cxt))) {
 		warnx(_("bad usage"));
@@ -905,7 +906,7 @@ int main(int argc, char **argv)
 		if (mnt_context_is_restricted(cxt) &&
 		    mnt_context_get_source(cxt) &&
 		    mnt_context_get_target(cxt))
-			exit_non_root(NULL);
+			suid_drop(cxt);
 
 	} else if (argc == 1 && (!mnt_context_get_source(cxt) ||
 				 !mnt_context_get_target(cxt))) {
@@ -933,7 +934,7 @@ int main(int argc, char **argv)
 		if (mnt_context_is_restricted(cxt) &&
 		    mnt_context_get_source(cxt) &&
 		    mnt_context_get_target(cxt))
-			exit_non_root(NULL);
+			suid_drop(cxt);
 
 	} else if (argc == 2 && !mnt_context_get_source(cxt)
 			     && !mnt_context_get_target(cxt)) {
@@ -941,7 +942,7 @@ int main(int argc, char **argv)
 		 * D) mount <source> <target>
 		 */
 		if (mnt_context_is_restricted(cxt))
-			exit_non_root(NULL);
+			suid_drop(cxt);
 
 		mnt_context_set_source(cxt, argv[0]);
 		mnt_context_set_target(cxt, argv[1]);
@@ -963,6 +964,14 @@ int main(int argc, char **argv)
 		mnt_context_set_optsmode(cxt, MNT_OMODE_NOTAB);
 
 	rc = mnt_context_mount(cxt);
+
+	if (rc == -EPERM
+	    && mnt_context_is_restricted(cxt)
+	    && !mnt_context_syscall_called(cxt)) {
+		/* Try it again without permissions */
+		suid_drop(cxt);
+		rc = mnt_context_mount(cxt);
+	}
 	rc = mk_exit_code(cxt, rc);
 
 	if (rc == MNT_EX_SUCCESS && mnt_context_is_verbose(cxt))
diff --git a/sys-utils/umount.8 b/sys-utils/umount.8
index f94d2f41f..9bec521e7 100644
--- a/sys-utils/umount.8
+++ b/sys-utils/umount.8
@@ -190,6 +190,25 @@ Display version information and exit.
 .TP
 .BR \-h , " \-\-help"
 Display help text and exit.
+.SH "NON-SUPERUSER UMOUNTS"
+Normally, only the superuser can umount filesystems.
+However, when
+.I fstab
+contains the
+.B user
+option on a line, anybody can umount the corresponding filesystem.  For more details see
+.BR mount (8)
+man page.
+.PP
+Since version 2.34 \fBumount\fR command allows to perform umount operation also
+for fuse filesystems if kernel mount table contains user's ID.  In this case fstab
+user= mount option is not required.
+.PP
+Since version 2.35 \fBumount\fR command does not exit when user permissions are
+inadequate by internal libmount security rules.  It drops suid permissions
+and continue as regular non-root user. It allows to support use-cases where
+root permissions are not necessary (e.g. fuse filesystems, user namespaces,
+etc).
 .SH "LOOP DEVICE"
 The
 .B umount
diff --git a/sys-utils/umount.c b/sys-utils/umount.c
index 397e0ebfc..74d87d671 100644
--- a/sys-utils/umount.c
+++ b/sys-utils/umount.c
@@ -112,24 +112,24 @@ static void __attribute__((__noreturn__)) usage(void)
 	exit(MNT_EX_SUCCESS);
 }
 
-static void __attribute__((__noreturn__)) exit_non_root(const char *option)
+static void suid_drop(struct libmnt_context *cxt)
 {
 	const uid_t ruid = getuid();
 	const uid_t euid = geteuid();
 
-	if (ruid == 0 && euid != 0) {
-		/* user is root, but setuid to non-root */
-		if (option)
-			errx(MNT_EX_USAGE,
-				_("only root can use \"--%s\" option "
-				 "(effective UID is %u)"),
-					option, euid);
-		errx(MNT_EX_USAGE, _("only root can do that "
-				 "(effective UID is %u)"), euid);
+	if (ruid != 0 && euid == 0) {
+		if (setgid(getgid()) < 0)
+			err(MNT_EX_FAIL, _("setgid() failed"));
+
+		if (setuid(getuid()) < 0)
+			err(MNT_EX_FAIL, _("setuid() failed"));
 	}
-	if (option)
-		errx(MNT_EX_USAGE, _("only root can use \"--%s\" option"), option);
-	errx(MNT_EX_USAGE, _("only root can do that"));
+
+	/* be paranoid and check it, setuid(0) has to fail */
+	if (ruid != 0 && setuid(0) == 0)
+		errx(MNT_EX_FAIL, _("drop permissions failed."));
+
+	mnt_context_force_unrestricted(cxt);
 }
 
 static void success_message(struct libmnt_context *cxt)
@@ -220,6 +220,15 @@ static int umount_one(struct libmnt_context *cxt, const char *spec)
 		err(MNT_EX_SYSERR, _("failed to set umount target"));
 
 	rc = mnt_context_umount(cxt);
+
+	if (rc == -EPERM
+	    && mnt_context_is_restricted(cxt)
+	    && !mnt_context_syscall_called(cxt)) {
+		/* Failed somewhere in libmount, drop perms and try it again */
+		suid_drop(cxt);
+		rc = mnt_context_umount(cxt);
+	}
+
 	rc = mk_exit_code(cxt, rc);
 
 	if (rc == MNT_EX_SUCCESS && mnt_context_is_verbose(cxt))
@@ -494,7 +503,7 @@ int main(int argc, char **argv)
 
 		/* only few options are allowed for non-root users */
 		if (mnt_context_is_restricted(cxt) && !strchr("hdilqVv", c))
-			exit_non_root(option_to_longopt(c, longopts));
+			suid_drop(cxt);
 
 		err_exclusive_options(c, longopts, excl, excl_st);
 
-- 
2.21.0


             reply index

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-19 14:34 Karel Zak [this message]
2019-12-03 14:43 ` Karel Zak

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191119143408.b6pcbhtsiubxdcob@10.255.255.10 \
    --to=kzak@redhat.com \
    --cc=util-linux@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Util-Linux Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/util-linux/0 util-linux/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 util-linux util-linux/ https://lore.kernel.org/util-linux \
		util-linux@vger.kernel.org
	public-inbox-index util-linux

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.util-linux


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git