On Wednesday, F5 announced the release of patches addressing 22 security vulnerabilities in NGINX, including five critical issues that could enable unauthenticated remote code execution (RCE). These vulnerabilities, collectively termed NGINX2Root, pose significant risks to affected systems.
Details of the NGINX2Root Vulnerabilities
The critical issues, identified as CVE-2026-113425, CVE-2026-121810, CVE-2026-123007, CVE-2026-123633, and CVE-2026-123634, have been assigned a CVSS score of 9.2. CVE-2026-113425 specifically affects NGINX instances functioning as mail proxy servers, which can be exploited through specially crafted mail sessions. Another flaw, CVE-2026-121810, can be triggered on servers using HTTP/3, causing a use-after-free error in the NGINX worker process.
CVE-2026-123634 targets proxy-configured servers, where exploitation occurs if variables are included in the proxied server name and a subrequest is made via Server Side Includes (SSI) or mirror directives. The remaining vulnerabilities, CVE-2026-123007 and CVE-2026-123633, necessitate HTTP/2 and can lead to buffer overflow and use-after-free errors, respectively.
Impact and Exploitation Insights
Research by Ikotas Labs, Metnir, Rapid7, Wiz, and Xint reveals that these vulnerabilities, if exploited, allow attackers to execute commands with the same privileges as the NGINX worker process, potentially compromising the entire server. Notably, CVE-2026-123007 has been demonstrated to execute code on NGINX serving static files without backend applications or third-party modules.
Xint has highlighted the potential of chaining CVE-2026-123007 with a Linux kernel vulnerability to exit a Docker container and obtain root access on the host system. While F5 notes that code execution depends on ASLR being disabled or bypassed, Wiz points out that CVE-2026-123007 and CVE-2026-123633 can bypass ASLR independently. Rapid7’s public exploit further demonstrates ASLR bypass by leaking memory addresses from server responses.
Proof-of-Concept Releases and Mitigation Measures
Rapid7 has released proof-of-concept (PoC) code targeting CVE-2026-123007, indicating that successful RCE exploitation requires substantial network traffic—approximately 128 GB of HTTP/2 data. Depending on network conditions, the time needed for exploitation can range from two minutes to an hour. On slower connections, NGINX’s standard one-hour connection duration can hinder exploitation efforts.
Censys, an internet intelligence provider, has confirmed the feasibility of reproducing CVE-2026-123007’s effects using cleartext HTTP/2 without the high traffic volume noted by Rapid7. Their research highlights that around 9 million NGINX services run affected versions with HTTP/2 enabled, with another 103 million services not disclosing version information.
F5’s patches for these vulnerabilities are available in NGINX Open Source versions 1.30.7 and 1.31.9, along with NGINX Plus versions 37.0.8.1 and 37.3.0.1. The updates also address additional flaws that could result in SSL/TLS connection termination, certificate verification bypass, segmentation faults, and denial-of-service (DoS) scenarios.
The security updates extend to 12 BIG-IP vulnerabilities, including critical and high-severity issues potentially leading to RCE, DoS, file manipulation, and token lifetime manipulation. F5 has not reported any active exploitation of these vulnerabilities in the wild. Further details are available in F5’s out-of-band security notification.
