On this page
Java4Shell is the nickname for CVE-2026-114938, a memory-corruption flaw in the Base64 decoder of OpenJDK, disclosed by security firm Calif on 17 November 2026. Oracle shipped fixed builds the same day, including 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1. Calif says it can lead to remote code execution on x86-64 CPUs with AVX-512 VBMI.
What is Java4Shell?
Java4Shell is the informal name Calif gave to a bug in the x86-64 intrinsic implementation of the Base64 decoder in OpenJDK. An intrinsic is a hardware-accelerated shortcut the Java virtual machine uses in place of the normal Java code. According to JFrog's summary, the flaw is an out-of-bounds heap write that the reporter says can lead to heap disclosure and remote code execution.
The write-up reached the top of Hacker News on 17 November 2026 with more than 1,000 points. ## Which versions are affected, and which are fixed?
Calif states that Java4Shell affects OpenJDK from version 17.0.6 or newer. Tenable's FAQ reports that the OpenJDK Vulnerability Group lists the 17, 21, 25 and 27 release lines, and that Oracle fixed them in its November 2026 Critical Security Patch Update on 17 November 2026.
| Release line | Affected (Oracle build listed) | Oracle fixed build |
|---|---|---|
| Java 17 | 17.0.21.0.3 (17.0.6 onwards per Calif) | 17.0.21.1 |
| Java 21 | 21.0.13.0.3 | 21.0.13.1 |
| Java 25 | 25.0.5.0.3 | 25.0.5.1 |
| Java 27 | 27.0.1.0.3 | 27.0.1.1 |
| Java 8 Enterprise Performance Pack | 8u511b33-perf | 8u513-perf |
Upwind's analysis adds that standard Java 8 has no vulnerable decoder, Java 11 is not affected, and Java 17.0.0 to 17.0.5 is also not affected. Oracle's fixed builds carry a ".1" suffix.
Other OpenJDK-based distributions are a separate matter. Calif lists Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu, BellSoft Liberica, SAP SapMachine and Red Hat builds among those derived from OpenJDK. Upwind says fixes from vendors other than Oracle are still rolling out, so check each vendor's own advisory for the exact fixed version.
How serious is the CVE-2026-114938 score?
Two scores are circulating. Calif gives a CVSS 4.0 base score of 9.2 (critical). Tenable reports that Oracle rates it 8.1 under CVSS 3.1, with high attack complexity. Oracle also says exploitation requires an application to pass untrusted data to the affected Java APIs, according to Tenable.
Who is exposed?
Calif says only x86-64 processors that support AVX-512 VBMI are affected, and that ARM64 is not. It names Intel Xeon Ice Lake, Sapphire Rapids, Emerald Rapids and Granite Rapids, and AMD EPYC and Ryzen Zen 4 or newer, as examples. JFrog reports that i386 and i686 are not affected either.
Calif describes the attacker as remote and unauthenticated, and says client-side cases are possible where a Java program connects to a malicious server. These are the discoverer's claims. Calif says it has demonstrated remote code execution in several widely used Java applications in default configurations but is withholding those product-specific chains until users can upgrade, so they cannot be independently verified yet.
Is it being exploited?
Not as far as the sources we opened report. Tenable says it is not aware of exploitation reports and that the flaw is not in CISA's Known Exploited Vulnerabilities catalogue. Upwind also reports no exploitation in the wild and no KEV listing. Tenable adds that Calif published a proof of concept that triggers the bug without taking control of a system, and that one researcher says he has a working exploit he has not released.
What should administrators do?
The primary fix is to upgrade to a patched build. JFrog's advice is to inventory Java deployments first, prioritising internet-facing x86-64 systems.
If an upgrade cannot happen straight away, Calif suggests two interim options: starting the JVM with the flag that disables the Base64 intrinsic, or disabling AVX-512 with a UseAVX setting. Calif says the second applies only to x86-64 hosts, because an ARM JVM will fail to start with it. Tenable points out that neither Oracle nor the OpenJDK Vulnerability Group mentions these options, so treat them as the discoverer's workaround, and expect some performance cost.
Analysis: what stands out
This section is our own reading of the published facts.
First, the hardware condition narrows exposure but should not give false comfort. AVX-512 VBMI is common on recent server-class chips, and cloud instance types often hide which CPU a workload lands on.
Second, the flaw sits in a core library, so the question for most teams is not whether they "use Base64" but whether any application in their estate runs an affected Java build and handles untrusted input. That makes an inventory the hard part.
Third, early coverage has been messy. Upwind lists several circulating claims it says are wrong, including that only Java 17 is affected. Official pages from OpenJDK and Oracle returned access errors when we tried to open them, so the version details here come from Calif, Tenable, JFrog and Upwind, and should be checked against the vendors' own advisories.
Similar patch-now stories have appeared recently, such as the Apache HTTP Server critical flaw fixed in 2.4.71 and the ScreenConnect extension flaw. For a company that has just been through an incident, our report on the Maersk legacy-system breach is a reminder of how older systems linger.
Sources
FAQ
Frequently asked questions
Java4Shell is the nickname for CVE-2026-114938, a memory-corruption flaw in the x86-64 Base64 decoder intrinsic in OpenJDK. Calif, which reported it, says it can lead to heap disclosure and remote code execution.
Oracle's fixed builds are 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1, plus 8u513-perf for the Java 8 Enterprise Performance Pack, according to Tenable. Other distributions have their own fixed versions, so check each vendor.
Per Upwind, standard Java 8 and Java 11 are not affected. Oracle lists the Java 8 Performance Pack build 8u511b33-perf as affected.
The sources we opened report no known exploitation and no CISA KEV listing as of 18 November 2026. Calif says it is withholding product-specific exploit chains until users can upgrade.
Calif and JFrog say ARM64 is not affected. The flaw needs an x86-64 CPU with AVX-512 VBMI support.
Spotted an error? Tell us. Read how we report and correct stories in our editorial policy.
Written by
Mainstream Tech
Newsroom
News and analysis on technology, AI, business, politics and international affairs, from our newsroom in London.
More from Mainstream




