---
schema: 1
kind: vulnerability
title: "Java4Shell, CVE-2026-114938, OpenJDK 17.0.6 and later and Oracle Java SE on x86-64 CPUs with AVX-512 VBMI: a heap overflow in the Base64 decoder that its discoverer chains into pre-authentication code execution in Keycloak, Jira, Confluence, Artifactory, Jenkins and PingFederate; fixed in 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1, no exploitation reported (CVSS 3.1 8.1)"
headline: "Java4Shell: OpenJDK Base64 heap overflow chained to pre-auth RCE in Jira and Keycloak; update Java, no attacks yet"
summary: >
  Oracle's Critical Security Patch Update and the OpenJDK advisory of 2026-11-17 fix CVE-2026-114938, an out-of-bounds write in the AVX-512 Base64 decoder that Calif, its discoverer, scores CVSS 4.0 9.2 and says it turned into pre-authentication code execution in the default configurations of Keycloak, Jira, Confluence, Artifactory, Jenkins and PingFederate; Oracle scores it CVSS 3.1 8.1 with high attack complexity. Only x86-64 hosts with AVX-512 VBMI are affected, a trigger proof of concept is public, Calif withholds its product chains, a second researcher claims a working exploit with no release stated, and no source reports attacks. NCSC.ch and BACS warn operators to update internet-reachable Java applications and identity services first.
discovered_at: "2026-11-18T05:03:30Z"
updated_at: null
event_date: "2026-11-17"
run_id: 2026-11-18T0356Z-intel
priority: high
immediate_action: null
tags: [vulnerabilities, rce, pre-auth, poc-public, patch-available]
regions: [global, switzerland]
sectors: [technology]
entities: []
techniques: [T1190, T1203]
affected_products: ["OpenJDK", "Oracle Java SE", "Oracle GraalVM for JDK"]
cves:
  - id: CVE-2026-114938
    cvss: "8.1"
    epss: null
    type: memory-corruption
    vector: zero-click
    auth: pre-auth
    status: [poc-public, patch-available]
    affected: "OpenJDK 17.0.6 and later, 21, 25 and 27 up to 17.0.21.0.3, 21.0.13.0.3, 25.0.5.0.3 and 27.0.1.0.3 on x86-64 with AVX-512 VBMI; Oracle Java SE 8u511b33-perf"
    fixed: "Oracle builds 17.0.21.1, 21.0.13.1, 25.0.5.1, 27.0.1.1 and 8u513-perf"
sources:
  - url: "https://www.oracle.com/security-alerts/cspunov2026.html"
    publisher: "Oracle Critical Security Patch Update"
    date: "2026-11-17"
    role: primary
  - url: "https://openjdk.org/groups/vulnerability/advisories/2026-11-17"
    publisher: "OpenJDK Vulnerability Group"
    date: "2026-11-17"
    role: primary
  - url: "https://calif.io/java4shell"
    publisher: "Calif (discoverer)"
    date: "2026-11-17"
    role: primary
  - url: "https://security-hub.ncsc.admin.ch/#/posts/13198"
    publisher: "NCSC.ch Cyber Security Hub"
    date: "2026-11-17"
    role: corroborating
  - url: "https://www.bacs.admin.ch/de/warnung-java-de"
    publisher: "BACS (NCSC.ch)"
    date: "2026-11-18"
    role: corroborating
  - url: "https://www.cisa.gov/news-events/alerts/2026/11/17/critical-java-vulnerability-requires-immediate-patching"
    publisher: "CISA"
    date: "2026-11-17"
    role: corroborating
  - url: "https://www.tenable.com/blog/java4shell-cve-2026-114938-frequently-asked-questions-about-openjdk-rce-vulnerability"
    publisher: "Tenable"
    date: "2026-11-17"
    role: corroborating
  - url: "https://blog.talosintelligence.com/java4shell/"
    publisher: "Cisco Talos"
    date: "2026-11-18"
    role: corroborating
  - url: "https://www.upwind.io/blog/java4shell-what-we-know-now-and-what-comes-next/"
    publisher: "Upwind"
    date: "2026-11-17"
    role: corroborating
  - url: "https://equixly.com/blog/2026/11/17/cve-2026-114938-pre-auth-heap-overflow-in-java-17-21-and-25/"
    publisher: "Equixly"
    date: "2026-11-17"
    role: corroborating
closed_sources: []
evidence:
  - quote: "Our initial testing demonstrated remote code execution in the following widely deployed Java applications, in their default configs."
    publisher: "Calif (discoverer)"
    source_url: "https://calif.io/java4shell"
  - quote: "Only the landing position of the out-of-bounds write is controllable."
    publisher: "Calif (discoverer)"
    source_url: "https://calif.io/java4shell"
  - quote: "We are holding back the application-specific exploits for now."
    publisher: "Calif (discoverer)"
    source_url: "https://calif.io/java4shell"
  - quote: "Because the vulnerable code is hardware-specific, Java4Shell also requires an x86-64 processor with AVX-512 VBMI support. ARM64 CPUs are not affected."
    publisher: "Calif (discoverer)"
    source_url: "https://calif.io/java4shell"
  - quote: "The published research demonstrates exploitation; it does not establish malicious exploitation in operational environments."
    publisher: "NCSC.ch Cyber Security Hub"
    source_url: "https://security-hub.ncsc.admin.ch/#/posts/13198"
  - quote: "A security update does not remove a compromise that has already happened."
    original: "Ein Sicherheits-Update beseitigt keine bereits erfolgte Kompromittierung."
    publisher: "BACS (NCSC.ch)"
    source_url: "https://www.bacs.admin.ch/de/warnung-java-de"
verification: multi-source
sourcing_note: >
  Oracle's CVE record scores CVSS 3.1 8.1 with high attack complexity and Calif scores CVSS 4.0 9.2; the Oracle score is the one recorded. The pre-authentication code-execution chains in six products are Calif's own demonstrations: no vendor statement and no released independent reproduction of them was found (Tenable relays a second researcher's claim of a working exploit with no release stated), and Upwind reproduced only the heap overflow. An early source-reading analysis (Equixly) expected a JVM crash; Calif's later write-up and Upwind's update say code execution was demonstrated.
confidence: medium
references: []
deep_dive: false
deep_dive_category: null
org_triage: null
classification:
  reliability: A
  credibility: 2
watchlist_hit: false
actions:
  - "On every x86-64 host with an AVX-512 VBMI processor that runs Java 17, 21, 25 or 27, update the JDK to 17.0.21.1, 21.0.13.1, 25.0.5.1 or 27.0.1.1 (Oracle builds; other distributions use their own build numbers) or 8u513-perf, then restart the Java processes and rebuild containers, internet-reachable applications and the identity, build and artifact servers Calif names (Keycloak, Jira, Confluence, Artifactory, Jenkins, PingFederate) first; where that waits, start the JVM with -XX:-UseBASE64Intrinsics, or with -XX:UseAVX=2 on x86-64 hosts only."
updates: []
migrated_from: null
---

Oracle's Critical Security Patch Update and the OpenJDK advisory of 2026-11-17 list CVE-2026-114938 in the hotspot compiler component, which Oracle rates CVSS 3.1 8.1 with high attack complexity ([Oracle, 2026-11-17](https://www.oracle.com/security-alerts/cspunov2026.html); [OpenJDK, 2026-11-17](https://openjdk.org/groups/vulnerability/advisories/2026-11-17)). Calif, which found it, says HotSpot's x86-64 Base64 intrinsic, which uses AVX-512 VBMI instructions, calculates a negative output length for a short padded input and that length becomes a full 64-byte write that sends 62 bytes past the end of the array; only the landing position of the write is controllable, a misplaced write can crash the JVM ([Calif, 2026-11-17](https://calif.io/java4shell)). It scores the flaw CVSS 4.0 9.2 and reports pre-authentication code execution in the default configurations of Keycloak, Jira, Confluence, Artifactory, Jenkins and PingFederate, authenticated execution in several further server products, and a client-side path in which a Java program that opens a TLS connection to a malicious server is compromised ([Calif, 2026-11-17](https://calif.io/java4shell)). Calif publishes a trigger proof of concept and withholds the product exploits ([Calif, 2026-11-17](https://calif.io/java4shell)). Upwind reproduced the overflow and says Basic-auth handling in Spring Security, Tomcat and Jetty 12, and Auth0's java-jwt, pass attacker-supplied tokens straight to the vulnerable decoder ([Upwind, 2026-11-17](https://www.upwind.io/blog/java4shell-what-we-know-now-and-what-comes-next/)). Tenable adds that Immersive's Kevin Breen claims a working exploit, adapted to each application and reaching the flaw before login through Tomcat's Basic authentication and java-jwt, with no release stated ([Tenable, 2026-11-17](https://www.tenable.com/blog/java4shell-cve-2026-114938-frequently-asked-questions-about-openjdk-rce-vulnerability)).

The flaw needs an x86-64 processor with AVX-512 VBMI and affects OpenJDK from 17.0.6 on, ARM64 hosts are not affected ([Calif, 2026-11-17](https://calif.io/java4shell)). CISA published an alert to patch immediately ([CISA, 2026-11-17](https://www.cisa.gov/news-events/alerts/2026/11/17/critical-java-vulnerability-requires-immediate-patching)), and the NCSC.ch Cyber Security Hub says the published research demonstrates exploitation without establishing malicious exploitation ([NCSC.ch, 2026-11-17](https://security-hub.ncsc.admin.ch/#/posts/13198)); Talos has seen no exploitation in the wild yet and expects probing ([Talos, 2026-11-18](https://blog.talosintelligence.com/java4shell/)).

**Exposure:** a Java 17, 21, 25 or 27 runtime, or the Java 8 performance build, at or below the affected builds ([Oracle, 2026-11-17](https://www.oracle.com/security-alerts/cspunov2026.html)), covering standalone runtimes, runtimes bundled in applications and appliances, and containers; the running build, not the installed package, decides ([NCSC.ch, 2026-11-17](https://security-hub.ncsc.admin.ch/#/posts/13198)). The CPU must report the avx512f, avx512bw, avx512vbmi and bmi2 flags, which means Intel Ice Lake to Granite Rapids and AMD Zen 4 or newer ([Upwind, 2026-11-17](https://www.upwind.io/blog/java4shell-what-we-know-now-and-what-comes-next/)). Pre-authentication reach depends on the application decoding attacker-controlled Base64 with the basic or URL decoder of java.util.Base64 before it authenticates the caller ([Upwind, 2026-11-17](https://www.upwind.io/blog/java4shell-what-we-know-now-and-what-comes-next/)).

**Detection:** unexplained hs_err_pid crash logs on internet-facing Java applications, where the log may not name Base64 because the crash comes later ([Equixly, 2026-11-17](https://equixly.com/blog/2026/11/17/cve-2026-114938-pre-auth-heap-overflow-in-java-17-21-and-25/)); as hunts that follow from Calif's code-execution chains, child processes of the application server and unusual administrative activity on identity, build and artifact platforms. In request and application logs, Upwind suggests Base64-alphabet tokens of 32 or more characters that end in = with a length not a multiple of 4, starting with Authorization headers, and bursts of IllegalArgumentException from java.util.Base64$Decoder or of restarts ([Upwind, 2026-11-17](https://www.upwind.io/blog/java4shell-what-we-know-now-and-what-comes-next/)); Talos publishes Snort rules for malformed Base64 in specific HTTP request fields ([Talos, 2026-11-18](https://blog.talosintelligence.com/java4shell/)). An update does not remove a compromise that has already happened ([BACS, 2026-11-18](https://www.bacs.admin.ch/de/warnung-java-de)).

**Defender takeaway:** update and restart, internet-reachable and identity services first. Where an update waits, Calif gives two JVM options, -XX:-UseBASE64Intrinsics, which disables the hardware-accelerated Base64 path, or -XX:UseAVX=2 on x86-64 hosts only, because OpenJDK on ARM does not recognize it and fails to start ([Calif, 2026-11-17](https://calif.io/java4shell)).
