Skip to content
World clockEU--:--UK--:--USA--:--CN--:--PLDEFRIT中文EN

portal about AI and technologyevents · analysis · interviews · technical background

Search
LIVE
›

Linux TCP flaw rated 9.8 gets a public exploit while Red Hat and Ubuntu race out patches

A Linux kernel use-after-free in IPv4 TCP, tracked as CVE-2026-98432 and rated 9.8, now has a working remote-root exploit. Fixes are only partly out: Red Hat published patches on 3 November, while Ubuntu's releases were still described as being for testing.

TechnologyExplainerRachel NwosuPublished: 3 October 20265 min readSources 8
Linux TCP flaw rated 9.8 gets a public exploit while Red Hat and Ubuntu race out patches

Red Hat shipped fixes for a 9.8-rated Linux kernel remote code execution flaw on 3 November. That came the same week researchers showed the bug can be turned into a working exploit. Ubuntu published its own kernel security updates around the same time. The flaw is CVE-2026-98432, a use-after-free in the IPv4 TCP stack.

What changes for anyone running Linux is simple. This is a network-reachable bug in the code that handles TCP connections. It is not a local privilege escalation that needs a logged-in user first. Servers exposed to the internet are the obvious target. Two things remain unclear. How many distributions have a fixed kernel in their stable repositories right now? And does the exploit in circulation work against every affected kernel version?

The exploit did not come from a traditional research team. Coverage on 2 November reported that an AI model produced a remote-root exploit for the flaw. That report attributes the work to a model rather than to named human researchers. The kernel developers have not independently confirmed it.

How the week unfolded

On 30 October, CVE-2026-98432 was published as a Linux kernel IPv4 TCP use-after-free issue. Two days later, on 2 November, reports said attackers had started scanning in the wild for the flaw. Those reports also said an AI model had produced a working remote-root exploit. By 3 November the picture had shifted from disclosure to incident response. Red Hat put out patches for the 9.8-rated TCP RCE, and Ubuntu's fixes were described as being released for testing. That sequence matters. A public exploit plus internet-wide scanning is the combination that turns a kernel CVE into a bad week for anyone who patches slowly. The 9.8 score, on the standard 10-point CVSS scale, puts the flaw in the critical band, close to the maximum.

A separate report from the Island outlet on 2 November said patches were not yet fully available. Red Hat was releasing and Ubuntu was still preparing. Red Hat's patches and Ubuntu's releases landed the following day.

The two accounts describe the same window from different angles rather than contradicting each other. They do disagree on how complete the fixes were at the moment each was written.

The disclosure problem underneath

This is not the first kernel security story of the autumn. The earlier ones explain why this one feels familiar. On 21 October, several Linux kernel vulnerabilities were accidentally posted to the public kernel mailing list. A day later, coverage reported that critical unpatched kernel flaws had been disclosed on a public list before fixes existed. The same pattern showed up with two CVSS 9.0 remote code execution flaws in KSMBD, the in-kernel SMB server. Those were disclosed on public mailing lists before the kernel fixes were ready, according to reporting on 22 October. KSMBD is the subsystem that lets a Linux box serve SMB shares. That makes it attractive to attackers looking for file servers.

In other words, the TCP flaw is arriving into a project that has already spent two weeks dealing with early disclosure. The public record so far does not state whether the CVE-2026-98432 timeline was coordinated or leaked. The kernel community has not said publicly how the bug reached the point of a working exploit.

The practical question for administrators is not whether to patch, but how fast they can get a fixed kernel onto internet-facing machines.

Downstream distributions move at different speeds

Distribution vendors are where this lands for most users. Debian's 13 "Trixie" kernel security update patched more than 1,300 CVEs, according to 9to5Linux on 2 October. CyberSecurityNews put the count at 1,313 flaws leading to denial of service and privilege escalation attacks. That is a bulk catch-up release, not a single-flaw fix. It shows how much kernel debt accumulates between updates. Ubuntu's kernel vulnerabilities also drew a separate advisory from the Hong Kong Computer Emergency Response Team Coordination Centre on 2 October. Intersect urged Cardano operators to patch CVE-2026-98432 on 31 October. That is a reminder that the people who need this fix fastest often do not run a general-purpose server fleet at all.

The spread of vendors in play is the reason a single kernel CVE produces a week of headlines. Those vendors include Red Hat, Ubuntu, Debian and the downstream projects that repackage their kernels. Each vendor builds, tests and signs its own kernel. Each one hits the release button on its own schedule.

What to do, and what nobody has said

  • Check whether your distribution has published a kernel package that names CVE-2026-98432, and whether it is in the stable or the testing channel.
  • Treat 2 November as the point when scanning began, not the point when the bug appeared.
  • If you run KSMBD, review exposure separately; the October flaws in that subsystem were disclosed before fixes existed.

Neither the kernel developers nor the major distributions have said how widely the flaw was exploited before patches shipped. No figure for affected hosts has been published.

The scans reported on 2 November show interest from attackers. They do not by themselves prove successful compromises.

The broader signal is about pace. A 9.8 kernel bug went from CVE publication on 30 October to a public exploit on 2 November to vendor patches on 3 November. That is fast by historical standards. It is still slower than the scanners.

Comments 0

Sources

8
  1. 01Compiling the Linux Kernel with GccrsEN
  2. 02A Leap in GNU/Linux Usage, Japan's Share in ParticularEN
  3. 03An Update on Orion for Linux and WindowsEN
  4. 04Some thoughts on security after ten years of qmail 1.0 (2007)EN
  5. 05PTXBench: Benchmarking and Adapting LLMs for GPU Kernel OptimizationEN
  6. 06Tobi Xu, a trilingual GNU/Linux distro for scientists and artistsEN
  7. 07Enterprise Web App and API security under your controlEN
  8. 08Underdog private personal on-device local AI for privacy and securityEN

All figures and quotations in this text come from the sources listed below.

Content prepared by the editorial team with AI assistance.

Rachel Nwosu

Rachel Nwosu

AI, models and technology

Rachel Nwosu covers AI, models and technology for FLASH24, working from public model documentation, benchmark releases and repository histories rather than press summaries, and she skips announcements that arrive without reproducible numbers. She checks training-data claims against dataset cards and reruns reported metrics where code is available. She spends much of her week interviewing researchers and engineers, tracking model launch calendars, and comparing vendor benchmarks with independent evaluations. Outside the desk she runs 3D printers, restores old computers, and tests how models learn from internet junk. She does not publish benchmark figures she cannot trace to a source.

Newsroom →

Comments

0
  1. No comments yet — be the first.

Write a comment

Comments are public. We do not publish abuse, spam or advertising.