Cri-o: checkpoint restore can be triggered from different namespaces
No exploitation in the wild has been reported. EPSS puts the chance of exploitation in the next 30 days at 1%. Fixed in 1.29.11, 1.30.8 and 1.31.3.
A vulnerability was found in CRI-O, where it can be requested to take a checkpoint archive of a container and later be asked to restore it. When it does that restoration, it attempts to restore the mounts from the restore archive instead of the pod request. As a result, the validations run on the pod spec, verifying that the pod has access to the mounts it specifies are not applicable to a restored container. This flaw allows a malicious user to trick CRI-O into restoring a pod that doesn't have access to host mounts. The user needs access to the kubelet or cri-o socket to call the restore endpoint and trigger the restore.
Affected
Vendors, products and packages with version ranges
| Vendor | Product | Versions | Fixed |
|---|---|---|---|
| Red Hat | Enterprise Linux | ||
| Red Hat | OpenShift Container Platform | ||
| <1.29.11, >=1.30.0 <1.30.8, >=1.31.0 <1.31.3 | 1.29.11, 1.30.8, 1.31.3 |
Timeline
Every event on this vulnerability
Related
Other vulnerabilities in Red Hat, strongest first
Embed
A live severity badge for an advisory, README or ticket