Unauthorized seccomp profile (GITA-SEC-018)

Seccomp is a Linux kernel feature developed to minimize kernel exposure to userland processes in such a way as to constrain them to four basic syscalls, only permitting reading, writing, returning and exiting operations. Recently, an extension of this feature was developed in order to enable more control over which syscalls to permit and to block, leveraging the Evolved Berkley Packet Filter (EBPF) to achieve this. The extension is called Seccomp BPF.

With Seccomp BPF, syscalls from a process can be matched against a specified filtering list, and in case it tries to invoke a blacklisted entry, a violation signal can be raised or the process terminated. This filtering list is denominated Seccomp profile.

Kubernetes can enable Seccomp BPF for containerized processes as long as the host's kernel supports it. But when using the Unconfined option, it will completely disable Seccomp. In case a specific profile has not been written for the process in question, Kubernetes carries a built-in filter list which suits well most use cases. To enable it, set the seccompProfile option to RuntimeDefault as the example below:

apiVersion: v1
kind: Pod
metadata:
  name: redis
spec:
  containers:
    - name: redis
      image: redis:latest
      securityContext:
        seccompProfile:
          type: RuntimeDefault

Moreover, it's possible to make every container execute with the default Seccomp filter list. To do this, start the Kubelet component with the flag --seccomp-default, or set the equivalent option on its configuration file.

For more information, visit the Kubernetes documentation