[RFC PATCH v2 14/14] device_cgroup: Allow mknod in non-initial userns if guarded

[Michael Weiß]

If a container manager restricts its unprivileged (user namespaced)
children by a device cgroup, it is not necessary to deny mknod()
anymore. Thus, user space applications may map devices on different
locations in the file system by using mknod() inside the container.

A use case for this, we also use in GyroidOS, is to run virsh for
VMs inside an unprivileged container. virsh creates device nodes,
e.g., "/var/run/libvirt/qemu/11-fgfg.dev/null" which currently fails
in a non-initial userns, even if a cgroup device white list with the
corresponding major, minor of /dev/null exists. Thus, in this case
the usual bind mounts or pre populated device nodes under /dev are
not sufficient.

To circumvent this limitation, allow mknod() by checking CAP_MKNOD
in the userns by implementing the security_inode_mknod_nscap(). The
hook implementation checks if the corresponding permission flag
BPF_DEVCG_ACC_MKNOD_UNS is set for the device in the bpf program.
To avoid to create unusable inodes in user space the hook also checks
SB_I_NODEV on the corresponding super block.

Further, the security_sb_alloc_userns() hook is implemented using
cgroup_bpf_current_enabled() to allow usage of device nodes on super
blocks mounted by a guarded task.

Signed-off-by: Michael Weiß <michael.weiss@xxxxxxxxxxxxxxxxxxx>
---
 security/device_cgroup/lsm.c | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/security/device_cgroup/lsm.c b/security/device_cgroup/lsm.c
index a963536d0a15..6bc984d9c9d1 100644
--- a/security/device_cgroup/lsm.c
+++ b/security/device_cgroup/lsm.c
@@ -66,10 +66,37 @@ static int devcg_inode_mknod(struct inode *dir, struct dentry *dentry,
 	return __devcg_inode_mknod(mode, dev, DEVCG_ACC_MKNOD);
 }
 
+#ifdef CONFIG_CGROUP_BPF
+static int devcg_sb_alloc_userns(struct super_block *sb)
+{
+	if (cgroup_bpf_current_enabled(CGROUP_DEVICE))
+		return 0;
+
+	return -EPERM;
+}
+
+static int devcg_inode_mknod_nscap(struct inode *dir, struct dentry *dentry,
+				       umode_t mode, dev_t dev)
+{
+	if (!cgroup_bpf_current_enabled(CGROUP_DEVICE))
+		return -EPERM;
+
+	// avoid to create unusable inodes in user space
+	if (dentry->d_sb->s_iflags & SB_I_NODEV)
+		return -EPERM;
+
+	return __devcg_inode_mknod(mode, dev, BPF_DEVCG_ACC_MKNOD_UNS);
+}
+#endif /* CONFIG_CGROUP_BPF */
+
 static struct security_hook_list devcg_hooks[] __ro_after_init = {
 	LSM_HOOK_INIT(inode_permission, devcg_inode_permission),
 	LSM_HOOK_INIT(inode_mknod, devcg_inode_mknod),
 	LSM_HOOK_INIT(dev_permission, devcg_dev_permission),
+#ifdef CONFIG_CGROUP_BPF
+	LSM_HOOK_INIT(sb_alloc_userns, devcg_sb_alloc_userns),
+	LSM_HOOK_INIT(inode_mknod_nscap, devcg_inode_mknod_nscap),
+#endif
 };
 
 static int __init devcgroup_init(void)
-- 
2.30.2