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

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

Search
LIVE
›

Linux kernel security patches land while gccrs admits it cannot yet build the kernel

Red Hat shipped fixes on 3 November for a Linux kernel TCP remote code execution flaw rated 9.8. The same week, the gccrs team told RustConf 2026 its Rust frontend for GCC still cannot compile the kernel.

TechnologyAnalysisGrace OkonkwoPublished: 3 October 20264 min readSources 9
Linux kernel security patches land while gccrs admits it cannot yet build the kernel
Red Hat shipped fixes for a 9.8-rated Linux kernel TCP remote code execution flaw on 3 November, the same week

Red Hat released patches for a Linux kernel TCP remote code execution flaw rated 9.8 on 3 November, according to coverage aggregated in reports. Ubuntu had already pushed fixes for testing. Reports of in-the-wild scanning against the bug, tracked as CVE-2026-98432, appeared from 2 November. The timing matters. The vulnerable component sits in the kernel's IPv4 TCP stack, which almost every networked Linux host depends on.

The disclosure trail has been messy. Reports on 21 and 22 October said several Linux kernel vulnerabilities had been posted accidentally to the public kernel mailing list. By 28 October, reports described the 9.8-scored TCP bug as affecting Ubuntu, Red Hat and other distributions. A separate 30 October post identified the flaw as an IPv4 TCP use-after-free. Vendors moved at different speeds. Red Hat's fixes and Ubuntu's testing releases both landed on 3 November. Earlier coverage on 2 November noted that patches were not yet fully available everywhere. That gap is the part worth watching. A 9.8 score on the CVSS scale usually means remote, unauthenticated access with little or no user interaction, and a kernel TCP flaw fits that profile.

The full advisory text is not public here, so the precise attack path is not spelled out. The sequence is clear: accidental public disclosure in October, scanning in the wild by early November, and fixes arriving at the start of this week.

What the vendors actually shipped

The patching picture is broader than one CVE. Debian's 13 "Trixie" kernel security update addressed more than 1,300 CVEs, according to 9to5Linux on 2 October. CyberSecurityNews put the same figure at 1,313 flaws and warned of denial-of-service and privilege escalation paths. The Hong Kong Computer Emergency Response Team Coordination Centre issued its own Ubuntu Linux kernel advisory the same day. Intersect separately urged Cardano operators to patch CVE-2026-98432, per a 31 October report. That shows the flaw reaching beyond general-purpose servers into stake pool infrastructure. Those numbers are not directly comparable. A distribution kernel update bundles fixes accumulated over time, not a single incident.

Still, the volume says something about maintenance load. Debian's update and Red Hat's 3 November release both land in the same window. Administrators running mixed fleets will deal with several advisory streams at once.

The compiler problem underneath

Security patching is only one half of the kernel's maintenance story. At RustConf 2026, Pierre-Emmanuel Patry and Arthur Cohen gave a talk titled "Compiling the Linux kernel with gccrs". Cohen opened by saying the title was a lie, according to LWN.net. Gccrs cannot currently compile the Linux kernel, for multiple reasons. One, Cohen said, is that he and Patry were laid off, though Open Source Security has since picked up their contracts. Patry's follow-up at Kangrejos the next week framed the nearer goal as getting gccrs to "miscompile the Linux kernel". That means producing a binary that can be tested at all, before anyone claims it works.

The two of them have made substantial progress, LWN reported, with work now focused on the core crate and then moving up the dependency graph to alloc and beyond.

The kernel's Rust code is split into crates including core, stdarch, alloc and pin_init, Patry explained. The older approach of compiling small snippets in isolation has reached its limits. That is why the project now concentrates on core first. GCC itself is nearly 40 years old and supports more backend architectures than rustc, plus compiler plugins many projects rely on. A working gccrs would let distributions build the whole kernel with one toolchain.

Background noise, and what to do

Elsewhere, Kagi said on 3 October it is ending development of Orion for Linux and Windows and open-sourcing both. The Linux beta no longer receives updates from Kagi after 2 October 2026. That is unrelated to kernel security but relevant to anyone standardising desktop tooling. Techrights, meanwhile, argued on 3 October that GNU/Linux usage has not collapsed. It cited 5.3% in Canada and 4.1% in Mexico over the past four weeks. That claim rests on its own reading of Cloudflare data rather than a published dataset.

For administrators, the practical sequence is unchanged. Identify hosts running affected kernels, apply the Red Hat or Ubuntu fixes now available, and watch for follow-up advisories. No report says whether a mainline fix has been merged upstream. Treat vendor packages as the source of truth until that is confirmed.

Comments 0

Sources

9
  1. 01Compiling the Linux Kernel with GccrsEN
  2. 02An Update on Orion for Linux and WindowsEN
  3. 03A Leap in GNU/Linux Usage, Japan's Share in ParticularEN
  4. 04In Case Anybody Still Believes Microsoft Lunduke Cares About GNU/LinuxEN
  5. 05PTXBench: Benchmarking and Adapting LLMs for GPU Kernel OptimizationEN
  6. 06Enterprise Web App and API security under your controlEN
  7. 07Tobi Xu – a trilingual GNU/Linux distro for scientists and artistsEN
  8. 08Some thoughts on security after ten years of qmail 1.0EN
  9. 09Underdog 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.

Grace Okonkwo

Grace Okonkwo

AI, models and technology

Grace Okonkwo covers AI, models and technology for FLASH24, working from primary sources such as model cards, API documentation and benchmark papers rather than vendor summaries. She checks training data provenance, evaluation conditions and reported scores against the underlying datasets before any figure reaches print. She interviews researchers and engineers directly, tracks release calendars from major labs, and compares successive model versions on the same tests. Her own self-hosting, home-network and documentation-reading habits feed straight into that desk, since she tests tools on her own hardware first. She does not publish benchmark claims without a reproducible method.

Newsroom →

Comments

0
  1. No comments yet — be the first.

Write a comment

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