Home / Blog / nginx2root: Five new nginx vulnerabilities
Tech News

nginx2root: Five new nginx vulnerabilities

F5 patched five critical nginx memory corruption flaws affecting 8.8 million servers, while edge proxies shield isolated origins from direct exploits.

By Dillip Chowdary • Nov 20, 2026 • Source: Cloudflare Blog

nginx2root: Five new nginx vulnerabilities
Image: Cloudflare Blog

At a glance

Affected product
nginx web and proxy server
Fixed versions
nginx 1.30.7 and 1.31.9
Disclosure date
November 18, 2026
Vulnerability severity
CVSS 3.1 8.1 and CVSS 4.0 9.2
Attack prerequisite
Bypassing Address Space Layout Randomization
Global exposure
8.8 million HTTP/2 services on Censys

F5 and the nginx project disclosed ten security vulnerabilities in nginx on November 18, 2026, including five memory corruption flaws designated as nginx2root. The vulnerabilities enable unauthenticated remote attackers to crash workers or execute code on servers, with secondary privilege escalation granting root access on Linux hosts. According to Cloudflare Blog's report, edge proxies shield downstream origin servers from the most dangerous flaws while administrators deploy official software updates.

This piece examines the memory corruption flaws, the companion Linux kernel flaw, and the specific configurations that remain exposed. It is written for systems engineers, network operators, and security architects managing self-hosted infrastructure who must assess exposure and prioritize patching. The analysis details edge proxy protections, vulnerability triggers, and required origin isolation steps.

01Edge runtime updates and origin protections

Cloudflare confirmed that its edge proxy infrastructure remained protected before public disclosure occurred. Most customer traffic at the edge routes through the FL2 proxy architecture, which engineers implemented in Rust. Engineering teams reviewed remaining legacy components built upon a customized nginx fork and completed global updates closing vulnerable code pathways prior to coordinated publication.

The vendor updates addressed five companion vulnerabilities beyond the memory corruption flaws, including CVE-2026-118899, which caused an infinite worker loop while reading request bodies over TLS 1.3. Proxied origin hostnames receive immediate shielding because edge nodes terminate incoming client connections for HTTP/2 and HTTP/3. Perimeter nodes inspect client traffic and establish distinct backend sessions, preventing malicious transport frames from reaching origin processes.

02Five memory corruption flaws across millions of servers

The nginx2root group comprises five distinct memory corruption vulnerabilities discovered by the Xint research team, each requiring specific runtime configurations. The vulnerabilities carry a High rating of 8.1 under CVSS 3.1 and a Critical rating of 9.2 under CVSS 4.0. Exploiting these flaws for arbitrary code execution requires bypassing Address Space Layout Randomization, which researchers achieved during testing by leaking target memory pointers.

CVEBug classAffected configurationFirst affected release
CVE-2026-123007Integer overflow heap read/writeHTTP/2 enabled1.9.5
CVE-2026-123633Use-after-freeHTTP/2 with proxy_pass and low worker_connections1.19.7
CVE-2026-121810Use-after-freeHTTP/3 QUIC with proxy_pass1.25.0
CVE-2026-123634Use-after-freeresolver with dynamic proxy_pass and subrequests0.6.18
CVE-2026-113425Use-after-freeMail proxy with proxy on and auth_http0.5.33

Internet scanning data collected by Xint on November 10 identified 56.7 million responsive nginx servers, with 98.9 percent of version-reporting systems falling within vulnerable ranges. Exposure metrics gathered by Censys on November 19 indicated approximately 8.8 million online services running affected releases with HTTP/2 active. Independent researchers from Rapid7, Wiz Research, and Ikotas Labs separately identified the primary integer overflow alongside F5 maintainers.

03Kernel SCTP socket bug and root escalation

Advertisement

Tech Pulse Daily

Get tomorrow's pulse first

Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.

The sixth vulnerability uncovered in the nginx2root investigation resides in the Linux operating system kernel rather than server application code. Researchers discovered a use-after-free condition inside the kernel SCTP send path that has persisted since Linux kernel version 4.15 in 2018. An attacker who gains unprivileged worker execution inside an nginx process can trigger this kernel flaw to obtain full root privileges.

The race condition occurs when an SCTP association migrates between sockets while a send operation waits, leaving the waiting process to release freed structures. Researchers demonstrated full host root compromise from within a restricted Docker container environment. Many standard distributions load the SCTP kernel module automatically upon socket request, exposing hosts that never intentionally configure SCTP networking.

04Edge protocol termination and frame filtering mechanisms

Triggering memory corruption through the HTTP/2 and HTTP/3 flaws requires delivering specialized protocol control sequences directly into the parsing engine. An attacker must transmit millions of WINDOW_UPDATE frames, manipulate SETTINGS frames to stall output buffers, and issue precisely timed RST_STREAM signals. When traffic routes through an intermediary edge proxy, those raw transport-layer sequences terminate at the reverse proxy boundary.

Perimeter proxies read incoming client requests, apply security rules, and establish independent upstream connections toward target web origins. Malicious transport frames, stream resets, and flow-control manipulations get consumed entirely by perimeter proxies and never traverse internal routes. Rapid7 confirmed this defensive behavior in technical documentation, noting that intervening proxies fail to retransmit attacker-crafted framing sequences required for memory corruption.

05Origin isolation methods and unshielded service configurations

Protection through perimeter filtering requires that attackers cannot discover or reach origin network addresses directly over the public Internet. If malicious actors bypass edge routing by connecting directly to origin IP addresses on web or mail ports, origin software parses raw frames without protection. Operators must restrict ingress networking exclusively to authorized proxy nodes through deliberate infrastructure isolation steps.

  1. Cloudflare Tunnel Establish an outbound connection from origin instances to edge nodes so that host firewalls block all inbound network ports.
  2. Firewall to Cloudflare IP ranges Restrict external packet filtering rules at the host or network perimeter to allow traffic exclusively from published proxy subnets.
  3. Authenticated Origin Pulls Enforce mutual transport layer security so origin web servers reject any client connection lacking an official proxy certificate.

Note: Server deployments that expose internal DNS resolver subrequests, unshielded mail proxy protocols, or publicly reachable origin IP addresses remain fully susceptible to exploitation. Cloudflare proxies do not terminate non-HTTP traffic, and attackers who bypass edge routing can reach origin processes directly. Infrastructure operators must install official software patches to guarantee complete protection.

Infrastructure teams must deploy fixed binaries, specifically nginx 1.30.7 or nginx 1.31.9, across all production environments to eliminate underlying software vulnerabilities. Systems that rely on mail routing directives or complex subrequests cannot rely on HTTP proxy layers for isolation. Operating system administrators should also monitor kernel security notices to patch the underlying SCTP privilege escalation pathway.

Frequently asked questions

Which nginx versions contain fixes for the nginx2root vulnerabilities?

Maintainers patched the vulnerabilities in nginx 1.30.7 and 1.31.9. Server operators must update their installations to these releases or vendor-provided backports and restart the service.

How does Cloudflare shield origin servers from CVE-2026-123007?

Cloudflare terminates client HTTP/2 connections at its edge nodes and negotiates separate connections to origin servers. Malicious protocol control frames sent by attackers are consumed at the edge and never reach backend instances.

Can an attacker achieve root access using the nginx2root flaws?

An attacker who compromises an unprivileged nginx worker can exploit a Linux kernel SCTP use-after-free bug present since kernel 4.15. This allows privilege escalation to root on the host system, even from inside a Docker container.

When are origin servers behind Cloudflare still exposed to attack?

Origins remain vulnerable if attackers bypass edge proxies to connect directly to origin IP addresses. Deployments running nginx mail proxy services or specific resolver subrequest configurations are also unshielded by HTTP proxies.

Sources

Dillip Chowdary

Author

Dillip Chowdary

Writes Tech Bytes coverage of AI, engineering, and the tools that actually ship. Editor of Tech Pulse Daily.

Related on Tech Bytes

Advertisement

5-min tech signal

Weekday briefing for engineers who skip the noise.

No spam · Unsubscribe anytime

Advertisement

✈️ CareerPilot

Your AI job-search copilot

Match your resume against live Ashby, Greenhouse & Lever openings — fit scores, job-specific resume optimization and email alerts.

Find matching jobs →

Free Tools

Browse all tools →