Linux BPF LSM: attaching to xfrm_decode_session turns packet classification into a kernel panic

Impact

BPF LSM programs could attach to the xfrm_decode_session hook. That hook is allowed to return an error, but security_skb_classify_flow() calls it from a void path and hits BUG_ON when it does - so a BPF LSM program returning non-zero panics the host on the next packet classified through that path. This matters on fleets that run eBPF security or networking agents on every node: a policy change rolled out cluster-wide is a cluster-wide panic, and GPU nodes with in-flight training jobs lose the whole job, not just a pod. The fix disables attachment to the hook outright rather than making the return value safe.

Who can reach it

Local and privileged - loading a BPF LSM program requires CAP_BPF/CAP_SYS_ADMIN and a kernel built with CONFIG_BPF_LSM. The realistic path is an operator-installed eBPF agent or a compromised privileged DaemonSet, not an ordinary tenant pod.

What to do

Fixed in the mainline Linux kernel BPF code by refusing attachment to the hook; take the distro kernel update and reboot the node. Until then, audit which BPF LSM programs your node agents attach and confirm none targets xfrm_decode_session. The record names no fixed release version, only the stable commits.

References

Related entries

• Linux kernel migrate_device: pte_pfn/pte_dirty called on non-present PTEs during device migrationCVE-2026-80686 · Linux kernel mm/migrate_device (migrate_vma_collect_pmd non-present PTE handling)Unscored
• Linux iommufd: vDEVICE alloc error path leaks the igroup mutex and deadlocks later vDEVICE operationsCVE-2026-80687 · Linux kernel iommufd/viommu (vDEVICE alloc ioctl igroup lock)Unscored
• Linux kernel idxd: double free of wq, engine and group structs on setup error and cleanup pathsCVE-2026-80698 · Linux kernel dmaengine idxd (Intel DSA/IAA wq, engine and group device setup)Unscored
• Linux kernel KVM arm64 vGIC: double-deactivate of nested IRQs trips an AmpereOne erratumCVE-2026-80699 · Linux kernel KVM/arm64 vGIC (nested-context IRQ deactivation)Unscored
• Linux kernel MGLRU: stale batch updates after memcg reparenting cause premature OOM killsCVE-2026-80719 · Linux kernel mm MGLRU (per-generation batch accounting across memcg reparenting)Unscored
• Microsoft Hyper-V: vmswitch fails to validate guest OID requestsCVE-2021-28476 · Microsoft Hyper-VCritical