Linux Kernel SCTP Flaw Enables Local Users to Escalate to Root and Escape Containers

Key Takeaways

• A use‑after‑free vulnerability (CVE‑2026-64564, nicknamed SCTPhantom) in the Linux SCTP networking stack has existed since kernel 2.6.25 (2008) and can be exploited to obtain full root privileges on a host.
• The flaw is local; it requires the SCTP protocol to be reachable and, for a container escape, the attacker must be able to open an SCTP socket inside the container.
• Patches were released in the stable kernel lines 7.1.6, 6.18.42, 6.12.101 and 6.6.148 on August 3, 2026; any older kernel with SCTP enabled should be updated immediately.
• Tencent’s Zhuque Lab demonstrated that, with the default seccomp profile and without CAP_NET_ADMIN or CAP_SYS_ADMIN, the bug can be used to break out of a container and gain root on the underlying host in six out of eight test attempts.
• The vulnerability stems from a mismatch between address validation and the path on which a delete operation is acted, allowing a freed SCTP path to be reused after a wildcard delete.
• While the flaw can lead to a kernel panic or denial‑of‑service, the lab’s exploit achieves privilege escalation; no public exploit code or known‑in‑the‑wild usage has been observed as of August 7, 2026.
• Because vendors may backport the fix without changing the upstream version string, administrators must consult their distribution’s security tracker to verify patch status.
• Disabling the SCTP kernel module removes the attack surface entirely when the protocol is not needed.

Overview of the SCTPhantom Vulnerability
The security issue resides in the Stream Control Transmission Protocol (SCTP) implementation within the Linux kernel. SCTP is a transport‑layer protocol designed to support multi‑homing and dynamic address reconfiguration, allowing a single association to span multiple network paths and to add or remove addresses while the association is active. The bug, tracked as CVE‑2026-64564 and dubbed SCTPhantom by its discoverers at Tencent Zhuque Lab, is a classic use‑after‑free condition that has persisted unchanged since the kernel 2.6.25 release in 2008, meaning every subsequent kernel version up to the present contained the flaw until the August 3, 2026 stable updates.

Technical Root Cause
The vulnerability arises from how the kernel processes SCTP ADDIP (Address Addition/Deletion) requests. When a packet contains an address, a delete request for that same address, and then a wildcard delete, the kernel first validates the delete against the packet’s source address. However, it subsequently performs the actual deletion on a path selected by a different address embedded inside the message. This mismatch causes the kernel to free the SCTP transport structure associated with the first address while still holding a reference to it. The subsequent wildcard delete then attempts to use the already‑freed pointer, leading to a use‑after‑free condition that can be leveraged to execute arbitrary kernel code with root privileges.

Patch Details and Mitigation
The fix, applied to the stable kernel branches 7.1.6, 6.18.42, 6.12.101 and 6.6.148 released on August 3, 2026, simply refuses a delete operation that targets the path currently being processed by the incoming message. By ensuring that the address used for validation matches the address used for the delete action, the patch eliminates the opportunity to free and reuse the same transport structure. Administrators running any older kernel with SCTP enabled should upgrade to one of these patched versions or apply the equivalent backport supplied by their Linux distribution.

Exploitability and Attack Surface
SCTPhantom is a local vulnerability: the attacker must already be able to send SCTP packets to the target system. This limits remote exploitation but does not diminish the risk in multi‑tenant environments such as containers, virtual machines, or shared hosting platforms where unprivileged users can open SCTP sockets. The flaw becomes especially dangerous when combined with a container escape, because gaining root on the host essentially compromises the entire underlying infrastructure.

Tencent’s Container Escape Demonstration
Tencent Zhuque Lab reported that they successfully turned SCTPhantom into a container escape. Their exploit initially required the sysctls net.sctp.addip_enable and net.sctp.addip_noauth_enable to be set, which would have necessitated the CAP_NET_ADMIN capability. However, the researchers later discovered a method to enable these features on a per‑socket basis, thereby bypassing the need for privileged capabilities. Using the default seccomp profile and without granting either CAP_NET_ADMIN or CAP_SYS_ADMIN, the lab’s exploit achieved root on the host in six out of eight attempts across several distributions (Debian 13, Ubuntu 24.04, Rocky Linux 9, RHEL 9, and OpenCloudOS). The write‑up did not disclose the specific container runtime used, noting that factors such as socket access restrictions, seccomp profiles, and user‑namespace policies can shift the exploit’s feasibility.

Broader Context and Related Findings
The discovery of SCTPhantom is part of a larger trend in which machine‑assisted research pipelines are uncovering long‑dormant kernel vulnerabilities. Tencent credits the find to Corvus AI, a multi‑agent research system designed to audit kernel code. This year, Corvus AI also contributed to the disclosure of GhostLock (another kernel‑level flaw) and coincided with the release of Zapscape, an unrelated KVM escape, with the same set of stable kernel updates patching both issues. The concurrent disclosure highlights how automated tools can resurface ancient bugs that have survived years of scrutiny.

Severity Scoring and Vendor Response
As of August 7, 2026, the vulnerability had not yet been assigned a Common Vulnerability Scoring System (CVSS) score by the National Vulnerability Database (NVD), nor had a weakness classification been added. Tencent’s internal assessment placed the flaw at 8.5 on CVSS v4.0, reflecting its high impact (potential full system compromise) tempered by the local attack vector and the requirement for SCTP to be enabled. Distributions have begun backporting the fix; however, because vendors may apply the patch without incrementing the upstream version string, administrators are advised to verify patch status via their distribution’s security advisories or changelogs rather than relying solely on the kernel version number.

Practical Recommendations for Administrators

1. Update Immediately – If you are running any kernel older than the patched stable releases (7.1.6, 6.18.42, 6.12.101, 6.6.148) and SCTP is loaded, upgrade to a patched kernel or apply the vendor‑provided backport.
2. Disable SCTP When Unused – If your workload does not require SCTP, blacklist the sctp module or set net.sctp.addip_enable=0 to remove the attack surface entirely.
3. Restrict Socket Access – Limit the ability of unprivileged users (including containerized processes) to open SCTP sockets via seccomp profiles, AppArmor/SELinux policies, or by dropping the CAP_NET_RAW capability where appropriate.
4. Monitor Vendor Advisories – Keep track of distribution‑specific CVE notices, as backporting may occur without a visible version change.
5. Review Container Hardening – Ensure that default seccomp profiles are not overly permissive, consider adding a rule that blocks socket(SOCK_SEQPACKET, IPPROTO_SCTP), and avoid granting unnecessary capabilities to containers.

By following these steps, organizations can mitigate the risk posed by SCTPhantom and reduce the likelihood of a local privilege escalation leading to full host compromise.