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-2026-98573 Overview
CVE-2026-98573 is a Linux kernel vulnerability in the networking subsystem. The flaw affects skb_checksum_setup_ipv6() in net/skbuff, where skb_maybe_pull_tail() subtracts skb_headlen(skb) from an unsigned max argument and passes the result to __pskb_pull_tail() as a signed int. The function does not verify that max is at least skb_headlen(skb), allowing an integer underflow when parsing IPv6 extension headers. The resulting negative length can be interpreted as a very large copy length by skb_copy_bits(), triggering out-of-bounds memory access during socket buffer processing.
Critical Impact
An integer underflow during IPv6 extension header parsing can lead to oversized memory copies in kernel socket buffer handling, potentially causing kernel memory corruption or denial of service.
Affected Products
- Linux kernel (networking subsystem, net/skbuff)
- Systems processing IPv6 traffic with extension headers
- Distributions shipping affected upstream kernel versions prior to the fix commits
Discovery Timeline
- 2026-10-18 - CVE-2026-98573 published to NVD
- 2026-10-19 - Last updated in NVD database
Technical Details for CVE-2026-98573
Vulnerability Analysis
The vulnerability resides in the Linux kernel's socket buffer (skb) handling code within the IPv6 checksum setup path. During IPv6 extension header parsing, skb_checksum_setup_ipv6() invokes skb_maybe_pull_tail() to ensure requested bytes are available in the linear portion of the socket buffer. The helper computes a delta by subtracting skb_headlen(skb) from an unsigned max argument, then forwards the result to __pskb_pull_tail() typed as a signed integer.
When the socket buffer already has a linear area larger than MAX_IPV6_HDR_LEN, the subtraction wraps around. The wrapped value becomes a large unsigned number that, when reinterpreted as a signed int, produces a negative delta. __pskb_pull_tail() then passes this negative length into skb_copy_bits(), where it is treated as an extremely large unsigned copy length. The resulting out-of-bounds memory operation can corrupt kernel memory or crash the kernel.
Root Cause
The root cause is an integer underflow driven by missing input validation. The code assumes max is always at least skb_headlen(skb), but this invariant does not hold during IPv6 extension header processing when the linear area already exceeds the maximum IPv6 header length. The implicit conversion between unsigned and signed integer types compounds the issue by hiding the underflow at a type-system level.
Attack Vector
The vulnerability is reachable through the network stack when the kernel processes IPv6 packets containing extension headers. A remote attacker who can deliver crafted IPv6 traffic to a vulnerable host may be able to trigger the underflow condition. The attack requires no authentication and operates against the kernel directly, since checksum setup runs during packet reception.
The upstream fix changes the three extension-header call sites to pass the requested length itself as the pull bound, ensuring the delta can no longer go negative. See the kernel commit at git.kernel.org for the patch implementation.
Detection Methods for CVE-2026-98573
Indicators of Compromise
- Unexpected kernel panics or oops messages referencing skb_copy_bits, __pskb_pull_tail, or skb_checksum_setup_ipv6 in dmesg output
- Kernel warnings tied to IPv6 traffic processing on hosts receiving external packets
- Abnormal network interface resets or driver recovery events correlated with inbound IPv6 flows
Detection Strategies
- Monitor kernel logs for crashes or stack traces originating in the IPv6 input path or socket buffer helpers
- Inspect inbound IPv6 traffic for unusual extension header chains, oversized headers, or malformed packet structures
- Correlate kernel instability events with recent IPv6 traffic patterns from untrusted sources
Monitoring Recommendations
- Enable kernel crash dump collection (kdump) to capture evidence if the underflow is triggered
- Audit kernel versions across the fleet to identify hosts running pre-patch builds
- Track EPSS and vendor advisories for status changes; current EPSS probability is 0.21%
How to Mitigate CVE-2026-98573
Immediate Actions Required
- Identify all Linux hosts running kernel versions that predate the fix commits and prioritize patching for internet-facing systems
- Apply the stable kernel update from your distribution vendor as soon as it becomes available
- Restrict exposure of untrusted IPv6 traffic to kernel network stacks where immediate patching is not possible
Patch Information
The upstream Linux kernel maintainers have merged the fix across multiple stable branches. Review the commit references published on git.kernel.org, including commit 3b4e0b0c, commit 6c28c31f, commit 718747af, commit 9c4e4321, commit a0ad0473, commit b2da5ed3, commit cfa643f2, and commit d7f47422. Monitor your distribution's security tracker for backported package releases.
Workarounds
- Where IPv6 is not required, disable IPv6 on affected hosts or restrict IPv6 extension header traffic at perimeter firewalls
- Deploy network ingress filters that drop IPv6 packets carrying unusual or deeply nested extension header chains
- Apply vendor-provided kernel live patches if available for your distribution
# Example: drop IPv6 packets with uncommon extension headers at the edge
# (Adjust to match your organization's IPv6 policy before deploying)
ip6tables -A INPUT -m ipv6header --header hop-by-hop,route,dst --soft -j DROP
# Verify installed kernel version against distribution advisories
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

