CVE-2026-99288, dubbed NoisyTCP, is a remote crash vulnerability in the Linux kernel that dates back to October 2015 and — according to researcher Hyunwoo Kim — went unnoticed until this year.
How NoisyTCP (CVE-2026-99288) causes a full-system crash
The bug resides in Linux's TCP networking code, where two parts of the kernel can "trip over each other" while processing incoming traffic, leaving one routine attempting to use memory that has already been freed. When that happens, Linux can panic and take the entire machine down. The confirmed impact so far is remote denial of service; there is no evidence yet that the flaw can be used to steal data or achieve general-purpose remote code execution.
Services and the attack surface: what can be hit straight out of the box
Hyunwoo Kim reports that common network-facing services such as Nginx, Apache, and SSH can be vulnerable without special configuration. An attacker does not need credentials: they only need access to an open TCP port on a system that processes incoming network traffic across multiple receive queues or CPUs. Because the flaw sits in the kernel rather than in individual applications, crashing a service inside a container can take down the host and any other workloads running on it.

Your scanner finds 4,000 vulns. Which 12 matter?
Nubivance is a Rapid7 Registered Partner delivering vulnerability management as a service - scanning, risk-based prioritization, and remediation follow-through across IT and OT.
Fix the backlogVirtual machines, containers and the difference in exploit timing
Kim's testing shows the bug is particularly effective against virtual machines. He was able to crash Nginx, Apache and SSH servers hosted in QEMU/KVM virtual machines within a timeframe of roughly five seconds to two minutes. Physical, bare-metal hardware resisted more: even after tailoring an exploit, Kim needed about 14 hours to crash a bare-metal server running Nginx.
Mitigation, patching status, and operational trade-offs
A fix was committed to the upstream Linux kernel on November 3. Administrators must check whether their distribution has shipped the updated kernel; Kim says he found and patched the bug this year and published details on November 15 once the disclosure embargo was lifted. Red Hat has rated the vulnerability Important and assigned a CVSS score of 7.5.
- Temporary workarounds Kim suggests include disabling TCP SYN cookies or restricting incoming packet processing to a single receive queue.
- Each workaround has drawbacks noted by Kim: disabling SYN cookies reduces protection against SYN flood attacks, and restricting packet processing to a single receive queue can hurt network performance.
- Kim also notes that disabling SYN cookies must be applied separately to containers that use their own network namespaces.
What this means for administrators, container operators, and VM hosts
Administrators: confirm whether your distribution has incorporated the patch committed upstream on November 3; until your vendor ships an updated kernel, weigh the operational impact of the suggested mitigations (reduced SYN protection or constrained packet processing).
Container operators: note that kernel-level crashes triggered from inside a container can ripple to the host. If you apply SYN-cookie changes, plan for namespace-specific settings because containers with their own network namespaces require separate configuration.
Virtual machine hosts: Kim's lab results show VMs on QEMU/KVM are more quickly taken down than bare metal. Hosts running many guest VMs should prioritize patching and assess the risk that a single remote crash could affect multiple tenants.
Hyunwoo Kim's discovery closes an 11-year gap from the bug's origin in October 2015 to public disclosure and upstream remediation in 2026. For now, the confirmed danger is remote denial of service without credentials; the researcher has demonstrated limited exploitation against VMs in a lab but has not produced a general-purpose remote code execution exploit. Administrators and operators must therefore balance the urgency of deploying upstream fixes with the trade-offs of interim mitigations while watching for distribution updates.




