Merge pull request #10210 from swagatbora90/preserve-flags-cherry-pick-1.7

dmcgowan · web-flow · commit 36bbb4de8ee4 · 2024-05-14T21:29:27.000-07:00
[release/1.7] Preserve CL_UNPRIVILEGED locked flags during remount of bind mounts
diff --git a/mount/mount_linux.go b/mount/mount_linux.go
@@ -26,6 +26,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/containerd/containerd/pkg/userns"
 	"golang.org/x/sys/unix"
 )
 
@@ -114,12 +115,56 @@ func (m *Mount) mount(target string) (err error) {
 
 	const broflags = unix.MS_BIND | unix.MS_RDONLY
 	if oflags&broflags == broflags {
+		// Preserve CL_UNPRIVILEGED "locked" flags of the
+		// bind mount target when we remount to make the bind readonly.
+		// This is necessary to ensure that
+		// bind-mounting "with options" will not fail with user namespaces, due to
+		// kernel restrictions that require user namespace mounts to preserve
+		// CL_UNPRIVILEGED locked flags.
+		var unprivFlags int
+		if userns.RunningInUserNS() {
+			unprivFlags, err = getUnprivilegedMountFlags(target)
+			if err != nil {
+				return err
+			}
+		}
 		// Remount the bind to apply read only.
-		return unix.Mount("", target, "", uintptr(oflags|unix.MS_REMOUNT), "")
+		return unix.Mount("", target, "", uintptr(oflags|unprivFlags|unix.MS_REMOUNT), "")
 	}
 	return nil
 }
 
+// Get the set of mount flags that are set on the mount that contains the given
+// path and are locked by CL_UNPRIVILEGED.
+//
+// From https://github.com/moby/moby/blob/v23.0.1/daemon/oci_linux.go#L430-L460
+func getUnprivilegedMountFlags(path string) (int, error) {
+	var statfs unix.Statfs_t
+	if err := unix.Statfs(path, &statfs); err != nil {
+		return 0, err
+	}
+
+	// The set of keys come from https://github.com/torvalds/linux/blob/v4.13/fs/namespace.c#L1034-L1048.
+	unprivilegedFlags := []int{
+		unix.MS_RDONLY,
+		unix.MS_NODEV,
+		unix.MS_NOEXEC,
+		unix.MS_NOSUID,
+		unix.MS_NOATIME,
+		unix.MS_RELATIME,
+		unix.MS_NODIRATIME,
+	}
+
+	var flags int
+	for flag := range unprivilegedFlags {
+		if int(statfs.Flags)&flag == flag {
+			flags |= flag
+		}
+	}
+
+	return flags, nil
+}
+
 // Unmount the provided mount path with the flags
 func Unmount(target string, flags int) error {
 	if err := unmount(target, flags); err != nil && err != unix.EINVAL {
