Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2024-26673 Overview
CVE-2024-26673 is a Linux kernel vulnerability in the netfilter subsystem, specifically within the nft_ct connection tracking module. The flaw stems from missing validation of layer 3 and layer 4 protocol numbers when creating custom expectations through nftables. A local attacker with CAP_NET_ADMIN privileges in a network namespace can exploit this to trigger kernel-side issues affecting confidentiality and availability. The upstream fix restricts the accepted protocol families to NFPROTO_IPV4, NFPROTO_IPV6, and NFPROTO_INET, and rejects layer 4 protocols that do not use port numbers.
Critical Impact
Local attackers holding network administration capabilities can leverage malformed nftables custom expectations to compromise kernel integrity and system availability.
Affected Products
- Linux kernel (multiple stable branches prior to the fixing commits)
- Linux kernel 6.8-rc1 and 6.8-rc2
- Debian GNU/Linux 10 (addressed via Debian LTS)
Discovery Timeline
- 2024-04-02 - CVE-2024-26673 published to the National Vulnerability Database (NVD)
- 2026-06-17 - Last updated in NVD database
Technical Details for CVE-2024-26673
Vulnerability Analysis
The vulnerability resides in the netfilter nft_ct module, which allows nftables rules to define custom connection tracking expectations. Expectations tell conntrack to anticipate related flows (such as the data channel of an FTP session) and are parameterized by a layer 3 family and a layer 4 protocol. Before the fix, the kernel accepted arbitrary values for both parameters when userspace created a custom expectation object.
Accepting families outside NFPROTO_IPV4, NFPROTO_IPV6, and NFPROTO_INET leaves nft_ct operating on protocol handlers that were never intended for expectation objects. Accepting layer 4 protocols without a port concept (for example, protocols where destination port is not meaningful) violates an internal invariant, because destination port is a mandatory attribute of the expectation. The result is unsafe operations inside the kernel conntrack code path.
Root Cause
The root cause is improper input validation [CWE-NVD-noinfo] in the custom expectations setup path of nft_ct. The code did not sanitize the layer 3 family or verify that the configured layer 4 protocol supports ports before registering the expectation.
Attack Vector
Exploitation requires local access with privileges sufficient to configure nftables, typically CAP_NET_ADMIN within a user or network namespace. On systems where unprivileged user namespaces are enabled, this capability is reachable by low-privileged local users. An attacker crafts an nftables rule that installs a custom ct expectation with an unsupported family or a portless layer 4 protocol, triggering the unsafe kernel path.
No public proof-of-concept exploit is listed, and the issue is not tracked in the CISA Known Exploited Vulnerabilities catalog. See the upstream commits such as kernel commit f549f34 and kernel commit 0f501da for the fix details.
Detection Methods for CVE-2024-26673
Indicators of Compromise
- Unexpected kernel oops, warnings, or crashes referencing nft_ct, nf_conntrack_expect, or expectation registration functions in dmesg or /var/log/kern.log.
- nftables rules containing ct expectation objects with non-IP families or layer 4 protocols that do not carry a destination port.
- Unprivileged processes creating network namespaces and loading nftables rulesets outside of normal administrative workflows.
Detection Strategies
- Audit running nftables rulesets with nft list ruleset on each host and review all ct expectation definitions for unusual protocol parameters.
- Enable kernel audit rules for the nf_tables netlink family and track NFT_MSG_NEWOBJ events creating expectation objects.
- Monitor for use of unshare, clone(CLONE_NEWNET|CLONE_NEWUSER), or setns by non-admin users followed by nftables configuration syscalls.
Monitoring Recommendations
- Forward kernel ring buffer and audit logs to a centralized analytics backend and alert on new netfilter warnings or crashes.
- Baseline normal nftables administrative activity and alert on configuration changes originating from non-admin accounts.
- Track kernel version inventory to confirm patched builds are rolled out across Linux fleets.
How to Mitigate CVE-2024-26673
Immediate Actions Required
- Apply the latest stable Linux kernel update from your distribution that includes the netfilter nft_ct fix.
- Debian 10 operators should install the kernel update referenced in the Debian LTS Announcement.
- Reboot affected systems after patching so the fixed kernel is loaded.
- Inventory hosts with unprivileged user namespaces enabled and prioritize them for patching.
Patch Information
The fix is distributed across multiple stable branches via commits f549f34, 0f501da, 38cc160, 65ee90e, 8059918, b775ced, and cfe3550. The patches reject non-IP families and portless layer 4 protocols when creating custom conntrack expectations.
Workarounds
- Disable unprivileged user namespaces where not required by setting kernel.unprivileged_userns_clone=0 (Debian/Ubuntu) or user.max_user_namespaces=0.
- Restrict CAP_NET_ADMIN to trusted administrators and avoid granting the capability to container workloads unless necessary.
- Remove or block nft_ct custom expectation usage via mandatory access control policies (SELinux, AppArmor) on hosts that cannot be patched immediately.
# Disable unprivileged user namespaces to reduce reachability of CAP_NET_ADMIN
sudo sysctl -w kernel.unprivileged_userns_clone=0
echo 'kernel.unprivileged_userns_clone=0' | sudo tee /etc/sysctl.d/99-harden-userns.conf
# Verify installed kernel version includes the netfilter fix
uname -r
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

