Skip to content
AIBRAINWORX / OpenSourceCoder.comLOCAL-FIRST. BUILT IN THE OPEN.
A.I. Cyber Security / Linux & Systems

CVE-2026-114938: Reported Pre-Authentication RCE in Java 17 and OpenJDK Base64 Decoder

Oracle patched CVE-2026-114938, a CWE-122 heap-based buffer overflow in OpenJDK HotSpot's AVX-512 VBMI Base64 decoder affecting Java 17, Oracle Java SE, and GraalVM.

Listen to this articleBrowser Voice
Ready
Official company logo or product image for cve-2026-114938-java-17-openjdk-base64-rce
Official image or organization logo from www.rapid7.com; identification only, no endorsement.
Event / context: 2026-11-17. Source check: October 28, 2026.

Oracle released emergency security patches on November 17, 2026, addressing CVE-2026-114938, a heap-based buffer overflow tracked as CWE-122 inside the hardware-accelerated Base64 decoder used by OpenJDK and downstream Java distributions. According to a Rapid7 Emergent Threat Response advisory, the vulnerability affects Java 17 and related runtimes including Oracle Java SE and Oracle GraalVM. Discovered by security researchers Calif and dubbed Java4Shell, the flaw can enable an unauthenticated remote attacker to crash a Java application or execute arbitrary code with the privileges of the underlying Java process.

Core Patch Announcement and Enterprise Exposure Scope

In its November security advisory, Oracle rated CVE-2026-114938 as High severity with a CVSS v3.1 score of 8.1 and noted high attack complexity. Despite that complexity rating, Rapid7 recommends that enterprise defenders identify and remediate affected systems, prioritizing Internet-exposed services, on an emergency basis outside routine patching schedules. That urgency reflects the broad footprint of OpenJDK-derived runtimes across web services, middleware, and backend infrastructure.

Rapid7 attributes the initial exploitation claims to Calif, whose disclosure states that the researchers demonstrated unauthenticated remote code execution against several popular Java-based applications. Calif also published a local proof of concept for CVE-2026-114938 designed to test whether a local JDK installation is vulnerable. While Rapid7 notes it is actively monitoring for additional developments, it advises administrators to consult Oracle's November security update notice and OpenJDK's vulnerability advisories for authoritative vendor release tracking.

HotSpot Architecture, AVX-512 VBMI, and Vulnerability Mechanics

At the engineering level, CVE-2026-114938 originates in a hardware-specific optimization within HotSpot, the Java virtual machine used across many OpenJDK distributions. To accelerate common encoding operations, affected OpenJDK runtimes enable an optimized Base64 decoding routine by default whenever the JVM detects compatible x86-64 processor extensions, specifically AVX-512 VBMI. When an application routes untrusted external input into that accelerated Base64 decoder, improper bounds handling triggers a CWE-122 heap-based buffer overflow.

Heap corruption inside a native JVM intrinsic carries severe stability and security consequences. Depending on memory layout and payload construction, overflowing the heap buffer can immediately terminate the Java virtual machine, causing denial of service, or overwrite adjacent control structures to achieve remote code execution under the operating system account running the Java process. Because the acceleration path is enabled automatically on supported silicon rather than requiring opt-in runtime flags, default OpenJDK, Oracle Java SE, and Oracle GraalVM configurations inherit the behavior automatically when deployed on capable hardware.

Pre-Authentication Attack Paths and Developer Workflow Impact

For software engineers and platform operators, the primary operational risk stems from how routinely Java applications decode Base64 strings prior to enforcing access controls. Rapid7 highlights HTTP Basic authentication as a concrete pre-authentication exposure pattern: a web server or API gateway often Base64-decodes incoming credential headers before verifying the extracted username and password. Any service that parses untrusted HTTP headers, API tokens, serialized payloads, or multipart fields through the JVM's standard Base64 decoder before authentication can expose the HotSpot flaw directly to unauthenticated network traffic.

This architectural reality complicates triage for DevOps and open-source maintainers managing containerized workloads and third-party Java dependencies. Simply inspecting application-level authentication logic is insufficient if Base64 decoding occurs inside upstream HTTP frameworks, ingress filters, or security libraries prior to user validation. Teams operating Java 17 and related OpenJDK builds must audit base container images, host JDK packages, and embedded GraalVM runtimes across both Internet-facing gateways and internal services reachable from untrusted networks. Where rapid patch rollout is constrained by testing pipelines, defenders can use Calif's published local proof of concept in controlled staging environments to verify whether a specific JDK build and CPU host combination exhibits the vulnerability.

Hardware Boundaries, Distribution Caveats, and Verification Limits

Crucially, exposure to CVE-2026-114938 is bounded by underlying processor capabilities. According to Red Hat's OpenJDK bulletin cited by Rapid7, there is no known exposure on ARM64 architectures or on x86-64 processors that lack AVX-512 VBMI support. For organizations operating heterogeneous cloud fleets or migrating workloads between x86-64 instance generations and ARM64 nodes, identical OpenJDK container images may behave differently depending on the host CPU features exposed to the HotSpot JVM.

Several evidentiary and operational caveats remain important as patching progresses. Although Calif reported achieving unauthenticated remote code execution against several popular Java applications, Oracle's CVSS v3.1 assessment emphasizes high attack complexity, indicating that reliable remote exploitation may depend on specific application memory conditions or heap states. Furthermore, Rapid7's initial advisory focuses on vendor patches and local verification rather than confirming widespread active exploitation in the wild. Downstream Linux distributions and open-source OpenJDK rebuilds often ship updates on staggered schedules, meaning operators must verify exact patched package versions per vendor rather than assuming uniform availability across every ecosystem channel.

Automated Vertex AI (gemini-4-hydrogen) primary-source report; reviewed 2026-11-19.

Follow your curiosity.

Pass the signal.