Skip to content

Image: tenable.com · rights & removal

Executive Summary

A vulnerability named Java4Shell, designated CVE-2026-114938, exists in the OpenJDK Base64 decoder functionality, which can lead to memory corruption and potential remote code execution without requiring prior authentication. The flaw stems from a performance shortcut used by Java on certain processors that allows specially crafted Base64 strings to write data beyond allocated memory. This vulnerability was disclosed on November 17, 2026, prompting Oracle to release security patches, including those in the November 2026 CSPU. Attackers can potentially read application memory ("JavaBleed") or execute code, depending on the environment and application context.
The vulnerability affects specific Java versions and processors, being tied to x86-64 processors supporting AVX-512 VBMI instructions. While Oracle provided fixed builds for several affected versions, other distributions like Amazon Corretto and Eclipse Temurin also released their own fixes. Mitigation involves specific Java startup options, such as `-XX:-UseBASE64Intrinsics`, which were recommended by some projects. Exploitation success depends on the presence of a processor supporting AVX-512 VBMI and the application's usage of Base64 decoding, with potential exploitation pathways differing based on the specific application being targeted, such as Keycloak or Jenkins.

Facts Only

* CVE-2026-114938 is an OpenJDK Base64 Decoder Memory Corruption Vulnerability (Java4Shell).
* The vulnerability has a CVSSv3.1 score of 8.1 and a CVSS 4.0 score of 9.2.
* Oracle released fixed builds in their November 2026 CSPU on November 17, 2026.
* The flaw involves a performance shortcut in Java that corrupts memory when decoding specially crafted Base64 strings.
* Exploitation requires the application to pass untrusted data to affected Java APIs via a web service.
* The vulnerability is specific to x86-64 processors supporting AVX-512 VBMI.
* Affected OpenJDK versions listed include Java 17, 21, 25, and 27 release lines.
* Specific Oracle builds affected include 8u511b33-perf, 17.0.21.0.3, 21.0.13.0.3, 25.0.5.0.3, and 27.0.1.0.3.
* Applications known to be potentially exploitable include Keycloak, Atlassian Jira, Jenkins, Confluence, JFrog Artifactory, PingFederate, Apache Kafka Broker, Apache Solr, Elasticsearch, Gerrit, and Spinnaker.
* Exploitation can occur before login in some applications or after login in others.
* Mitigations include startup options like `-XX:-UseBASE64Intrinsics` or `-XX:UseAVX=2`.
* Fixed Oracle builds for Java 17, 21, 25, and 27 are 17.0.21.1, 21.0.13.1, 25.0.5.1, and 27.0.1.1, respectively.

Full Take

The narrative surrounding Java4Shell moves from a specific technical finding to broad implications regarding the architecture of application security and patch distribution. The core tension lies between the immediate availability of fixes (Oracle's patches) and the fragmented reality of deployment across diverse Java distributions and hardware architectures, which complicates risk assessment for end-users. The fact that an attacker can achieve code execution without a login significantly elevates the severity, moving the focus from access control bypass to fundamental memory integrity, echoing historical patterns where remote code execution vulnerabilities unlock systemic access.
The pattern of releasing technical details alongside exploit availability—citing Calif's research and subsequent reports from GreyNoise and VulnCheck—demonstrates a convergence of security research and active exploitation channels. This ecosystem creates an accelerating feedback loop: researchers publish findings, exploit developers refine the attacks (as seen with Kevin Breen's work), and vendors respond with patching timelines. The existence of specific hardware requirements (AVX-512 VBMI) introduces a layer of systemic dependency; not all systems are equally at risk based solely on version number. This forces an examination of what constitutes "vulnerability" when physical constraints and software configuration intersect.
The implication for system operators is that defense must shift beyond simple patching to incorporate architectural awareness—understanding the interplay between JVM implementation, underlying CPU features, and application-layer data handling. The distribution of fixes across Oracle builds versus community distributions like Temurin forces an understanding that security posture is not just about the JDK version but the entire software supply chain and runtime environment.
Bridge questions: How can organizations effectively manage risk when disparate vendors release patches with varying definitions of affected scope, and what systematic process is needed to prioritize mitigation based on endpoint hardware capabilities? What are the long-term implications if vulnerabilities leverage specialized processor features like AVX-512 VBMI for exploitation?

From the original · Tenable Blog

Java4Shell is a flaw in how Java decodes Base64 data. Calif, which found it, says it can let attackers run code on widely used Java applications without logging in.
Read the full story at tenable.com

Sentinel unavailable

The automated check of the source article's wording did not complete, so no result is shown.

This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.

Java4Shell (CVE-2026-114938): Frequently asked questions about OpenJDK remote code execution vulnerability | Huntaegis