Java vendors issue Java4Shell fixes as projects update deployment guidance
Java4Shell fixes are rolling out across Java vendors. Check the runtime, CPU, containers and bundled applications before considering a deployment updated.
Java vendors are releasing fixes for Java4Shell, a flaw in OpenJDK that can let an unauthenticated attacker run code remotely under certain conditions. Fresh guidance from software projects shows why the work can extend beyond updating the version of Java installed on a server’s operating system.
The vulnerability, tracked as CVE-2026-114938, affects optimized Base64 decoding in the Java runtime. OpenJDK’s Nov. 17 security advisory lists the flaw, and Microsoft’s Nov. 18 announcement identifies fixed releases of its OpenJDK build. Other vendors are distributing their own updated packages through separate release channels.
For organizations using Java through business applications, the first question is which runtime each service actually starts. A machine can contain several installations. The Java program used by an administrator’s terminal can differ from the runtime inside a container, a packaged product or a service configured with its own executable path.
Three conditions help establish exposure
The original Calif research disclosure describes a memory-corruption problem in a CPU-optimized Base64 decoder. The vulnerable path depends on the runtime build and compatible x86-64 hardware with AVX-512 VBMI support. An application also has to reach the affected decoding operation through input an attacker can influence.
Eclipse Temurin’s advisory identifies affected JDK generations and states that JDK 8 and 11 are unaffected by this issue. It also describes the hardware requirement. A statement that a company uses Java is therefore only the beginning of the exposure check. The runtime, hardware and application path supply the information needed to make that assessment.
Teams can record three facts for each service. First, identify the executable and vendor package used by the running process. Second, check the host or virtual CPU capability relevant to the advisory. Third, establish whether the application handles untrusted data through the affected decoder. Where a vendor has issued product-specific instructions, those instructions should guide the assessment.
A version label on a package inventory helps, but it needs to match the running deployment. An old process may still be using the runtime with which it started. A replacement container image may have been downloaded without replacing the active container. Completing the update includes checking that the intended patched runtime is the one executing the workload.
Which releases vendors identify as fixed
Microsoft says its OpenJDK 17.0.21.1, 21.0.13.1 and 25.0.5.1 builds contain the fix. Temurin lists fixed versions 17.0.21.1+2, 21.0.13.1+2, 25.0.5.1+2 and 27.0.1.1+4. These numbers belong to the vendors and channels named in the advisories. Use the package identifier applicable to the distribution you deploy.
Temurin says binaries are published as testing completes, with container images following after the relevant binaries are available. That makes the download channel part of the verification. A release announcement can identify a fixed build while a particular architecture, installer or image tag is still progressing through publication.
Linux distributions can also deliver fixes through their own packages and backports. The Scala project’s guidance tells users to consult the distribution’s security tracker rather than relying solely on an upstream version string. The useful evidence is the vendor’s statement that the exact package contains the fix, together with confirmation that the application uses that package.
Why containers need their own update
Suppose a company runs a Java application inside a container image. Updating Java on the underlying host changes that host installation. The application may continue using the runtime packaged inside its image. The service owner must inspect the image’s contents and follow the application’s update and redeployment process.
A practical deployment record can include the application name, base image or product build, image digest, Java vendor and version, deployment time and verification result. That record lets the team distinguish an updated base image from a service still running the prior image. It also gives operations staff a precise reference if they need to review a restart or investigate a regression.
The same principle applies to software that bundles Java. A patch for the operating system’s package may have no effect on a product installed with a separate runtime. Check the product vendor’s advisory and supported update route. Replacing a bundled runtime independently can introduce compatibility questions that the vendor’s own package is designed to address.
Quarkus separates runtime and native deployments
Quarkus updated its status page at 09:00 UTC on Nov. 19. It distinguishes Java virtual-machine deployments from native executables and the images used to build them. Its guidance says native executables built with Mandrel are unaffected, while a JVM deployment depends on the JDK it uses.
The page also distinguishes builder images from runtime images. At that update, it reported the workaround applied to Mandrel and GraalVM Community Edition builder images, while UBI Java runtime images still needed the stated mitigation. These are different stages of a software pipeline. A corrected build environment does not establish that an already deployed runtime has been updated.
For a team maintaining both kinds of deployment, a single product-level status can hide that distinction. List the application’s execution mode alongside the relevant image and runtime. A native executable, a JVM service and a tool used only during the build each need the assessment appropriate to that role.
What to do while a fixed package is pending
Temurin and Microsoft recommend upgrading to a patched runtime. Where an immediate upgrade is unavailable, the advisories describe mitigations. Temurin identifies the JVM startup flag -XX:-UseBASE64Intrinsics to disable the affected optimized decoder. Follow the relevant vendor’s instructions and verify that the actual service has picked up the setting after restart.
A startup change has a scope. A flag added to one application’s configuration does not automatically reach other JVM processes on the machine, CI workers or separate containers. Record which services received it, who owns the remaining systems and when the temporary measure will be reviewed against a supported update.
The most useful completion check is a deployment-specific one. Confirm the patched package or documented workaround, restart or redeploy through the normal process, verify the running runtime, and exercise the application’s ordinary functions. Vendor release-status pages remain the place to check availability as packages and images progress.
This advisory concerns a specific Java flaw. Newsmerica’s report on Microsoft’s earlier outage covers a separate service incident. Follow our Tech & AI coverage for verified software and security developments as further vendor updates appear.
Trust & transparency
How we handled this story
Checked OpenJDK’s Nov. 17 CVE matrix, Microsoft and Temurin’s Nov. 18 releases, Calif’s disclosure, and Quarkus guidance updated Nov. 19 at 09:00 UTC. Preserved CPU and runtime conditions and vendor availability limits. No unsupported connection made to other outages or claims of exploitation scale.
Grounded in this reporting
Ask Newsmerica
Ask about this article. Answers quote the closest published passages and link back to their sources.
This tool retrieves article passages; it does not add outside facts.Keep exploring






