Skip to content
NewA calmer, faster ZeroHour

Linux kernel TCP flaw enables pre-auth remote corruption

Original title: NoisyTCP: Linux Kernel TCP Pre-auth Remote Corruption (CVE-2026-99288)

oss-security ()2
AI summary · x-ai/grok-4.7Summaries by Vani AI

After the distro embargo expired, the report describes a race between a stale ehash lookup and request-socket reuse.

Hyunwoo Kim published NoisyTCP, tracked as CVE-2026-99288, after the embargo agreed with linux-distros maintainers expired. The advisory describes pre-authentication remote memory corruption in the Linux kernel TCP stack. Triggering it requires a stale ehash lookup to run at the same time as freeing and reallocating an existing request socket. The snippet says the reporter patched the issue but gives no kernel versions.

Unconfirmed: The snippet does not say whether corruption leads to code execution or whether attacks are known.

Key facts
  • The report was posted November 15 after the embargo expired.
  • The reporter is Hyunwoo Kim (@v4bel).
  • No affected or fixed kernel versions are stated.
  • The text says the vulnerability was patched by its reporter.
Security assessment · AI, from the article

Impact: pre-authentication remote memory corruption

ProductAffected
Linux kernel—
What to do
  1. 1.Apply the distributor kernel fix for CVE-2026-99288.
Analyze CVE-2026-99288 with Vani AIImpact, affected versions, exploitation, mitigations and detection

Vulnerabilities mentioned

All
CVEVulnerabilityCVSSEPSSSignalsPublished
CVE-2026-99288
NULL deref in Linux TCP established lookups
——

From the original article

Posted by Hyunwoo Kim on Nov 15 The embargo agreed with the maintainers of linux-distros () vs openwall org has expired, so I am posting this report. # Abstract This report is a security advisory for NoisyTCP (CVE-2026-99288), the Linux Kernel TCP Pre-auth Remote Corruption vulnerability reported and patched by Hyunwoo Kim (@v4bel).

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.