A Java flaw called Java4Shell can let attackers without an account run their own code on Jira, Jenkins, Keycloak, and three other widely used applications, says Calif, the security research firm that disclosed it this week. The flaw is present in Java 17.0.6 and later, and the vulnerable code runs only on x86-64 Intel and AMD chips that support the AVX-512 VBMI feature. Tracked as CVE-2026-114938, the fix was released today by Oracle and the OpenJDK project, whose open-source code underpins Java builds from many vendors. Amazon has fixed its Corretto Java builds, but Red Hat is still preparing fixes for its own. The researchers are holding back their attack code for the six applications, and other labs have so far reproduced crashes and a memory leak but not the code execution. Canada's Cyber Center said this week that it was not aware of any attacks exploiting the flaw. Servers often decode Base64-encoded data, such as HTTP Basic login headers, SAML artifacts, and CSRF tokens, before checking whether a user is logged in. That is how an attacker without an account can exploit the flaw. The flaw is in a faster version of Java's Base64 decoder that runs on chips with AVX-512 VBMI. Java switches to it after compiling the decoder for speed, which it does for code that runs often. A malformed string that ends with an extra = makes this decoder write 62 or 63 bytes past the end of its output. Java rejects the string, but only after the write. The bytes written past the end are fixed, so an attacker controls only where they land. For that reason, early readings of the source code had pointed to a crash rather than code execution, as The Hacker News noted in an earlier report yesterday. The attacks worked without a login under the default settings of Keycloak, Atlassian Jira, Atlassian Confluence, JFrog Artifactory, Jenkins, and PingFederate, and with a login on Apache Kafka Broker, Apache Solr, Elasticsearch, Gerrit, and Spinnaker, Calif says in its technical write-up. The researchers' lists are incomplete, and they rate the flaw as 9.2 out of 10 under CVSS 4.0. They get past the fixed bytes by arranging the server's memory so that the overflow expands a neighboring array. They then use that array to write bytes of their choosing and a second array to read back from memory, a step they call JavaBleed. From there, the researchers say, the attack reaches the parts of Java that run code. There is no single exploit that works everywhere. Each target requires its own memory layout and timing, and some attacks are sensitive to network delays. A proxy that waits for complete requests, as nginx does by default, blocks some attacks but not all. A failed attempt can crash Java, and an attacker can often try again. Security company Upwind reproduced the memory corruption on an AWS server with an Intel Ice Lake chip, where it damaged the next object in memory and eventually crashed Java, according to its lab results. Blackpoint Cyber's Adversary Pursuit Group crashed a test Spring Boot application by sending about 28,300 malformed login requests over roughly 20 seconds, without ever logging in. The group also used the JavaBleed step in a local test to read a secret stored next to the overflow, which worked in all 30 runs. It did not reproduce remote code execution, its test results show. Oracle scores the flaw 8.1 out of 10 under CVSS 3.1, rating it hard to exploit but requiring no login. The description in its CVE record says, "Successful attacks of this vulnerability can result in takeover of Oracle Java SE." Red Hat also rates it as requiring no login, with a preliminary score of 9.8 out of 10. Amazon scores it 8.8 out of 10 and describes the risk in its Amazon Linux entry as "a JVM crash for unauthenticated users or, in certain application environments, code execution by an authenticated low-privileged user." Canada's Cyber Centre also describes the flaw as exploitable remotely without authentication in its government alert. The flaw was not in the KEV catalog of exploited vulnerabilities, kept by the US Cybersecurity and Infrastructure Security Agency (CISA), when The Hacker News checked it today. Who Is Exposed The vulnerable code is in Java 17.0.6 and later on the Java 17 line, and in Java 21, 25, and 27. Standard Java 8 and Java 11 do not contain it, but the Java SE 8 Performance Pack build 8u511b33-perf is affected. GraalVM for JDK 17.0.21.0.3, GraalVM for JDK 21.0.13.0.3, and GraalVM Enterprise Edition 21.3.20.0.3 are also affected. It runs only on x86-64 chips that support AVX-512 VBMI. These include Intel Xeon Ice Lake, Sapphire Rapids, Emerald Rapids, and Granite Rapids chips, as well as AMD Zen 4 or newer chips. ARM64 chips do not run it. Because the vulnerable code runs only on these chips, the same container image can run it on one server and not on another. On Linux, this command shows whether a server's chip has the four features the vulnerable code needs, and all four must appear in its output: grep -m1 '^flags' /proc/cpuinfo | tr ' ' '\n' | grep -E '^(avx512f|avx512bw|avx512vbmi|bmi2)$' The attacker also needs malformed Base64 to reach Java's standard decoder, through paths such as the login headers and tokens described earlier. Calif also says a Java program that opens a TLS connection to a malicious server can be compromised, even if the program runs no service of its own. The exploit for this came from another researcher, samczsun, and the researchers have not published how it reaches the decoder. Fixes and Workarounds The OpenJDK project's vulnerability advisory lists the flaw as fixed in this week's releases, and Oracle's security advisory lists the affected versions of its builds. For Oracle Java, the fixed builds are the new .1 releases in the table, all released this week. The release notes for 17.0.21.1 say it includes the update's security fixes, and the notes for 21.0.13.1, 25.0.5.1 and 27.0.1.1 say the same. For the Java 8 Performance Pack, the latest build is 8u513, available in both standard and Performance Pack versions. For GraalVM, the November security releases are GraalVM for JDK 17.0.21.1, GraalVM for JDK 21.0.13.1, and GraalVM Enterprise Edition 21.3.20.1. Java build Affected versions Fixed release Oracle Java SE 17 17.0.6 to 17.0.21.0.3 17.0.21.1 Oracle Java SE 21 21.0.13.0.3 and earlier 21.0.13.1 Oracle Java SE 25 25.0.5.0.3 and earlier 25.0.5.1 Oracle Java SE 27 27.0.1.0.3 and earlier 27.0.1.1 Amazon Corretto 17, 21, 25 and 27 Earlier builds, from 17.0.6 on for Corretto 17 17.0.21.1.2, 21.0.13.1.2, 25.0.5.1.2 and 27.0.1.1.4 Azul Zulu 17, 21, 25 and 27 Earlier builds, from 17.0.6 on for Zulu 17 17.70.201, 21.54.201, 25.38.201 and 27.30.201 (free builds) Red Hat Enterprise Linux 8, 9 and 10, and Red Hat build of OpenJDK 17, 21 and 25 17.0.6 and later, including 21 and 25 None yet The Corretto changelog for 17.0.21.1.2 lists the flaw as fixed, as do the changelogs for the new 21, 25, and 27 builds. Amazon Linux also marks its Corretto 17, 21, 25 and 27 packages as fixed. Azul's November update fixes the flaw in Zulu 17, 21, 25 and 27, according to its Zulu release notes. Red Hat has not yet released fixes for Red Hat Enterprise Linux 8, 9 and 10, or for its OpenJDK 17, 21 and 25 builds, according to its security bulletin. Containers based on its images need to be rebuilt once updated images are out. SAP released SapMachine 17.0.21.1, 21.0.13.1, 25.0.5.1, and 27.0.1.1 this week, but their release pages do not specify what they fix. Eclipse Temurin, the Microsoft Build of OpenJDK, and BellSoft Liberica had not listed fixed builds when The Hacker News checked today. Products that bundle their own Java runtime need an update from the product's maker. After updating, restart Java services, and rebuild and redeploy container images so the running process uses the new build. If you cannot update yet, start the JVM with one of these flags, and remove it once the fixed build is installed: -XX:-UseBASE64Intrinsics turns off only the faster Base64 code, which removes the vulnerable path and leaves other AVX-512 speedups in place. -XX:UseAVX=2 turns off AVX-512 entirely. Use it only on x86-64 servers, because Java on ARM64 does not recognize it and will fail to start. In Upwind's lab, both flags stopped the overflow on every affected version it tested. The company warns that another flag, -XX:DisableIntrinsic=_base64_decodeBlock, stops Java from starting when it is used on its own. To spot attempts, it suggests checking for the signs below, while warning that none of them proves a compromise and that their absence does not prove a server is safe: Base64 tokens of 32 characters or more that end in = and whose length is not a multiple of 4, especially in Authorization headers Bursts of IllegalArgumentException errors from java.util.Base64$Decoder Unexplained hs_err_pid crash logs or bursts of restarts A malformed input that triggers the overflow was posted publicly on September 9, in a proposed fix submitted to OpenJDK's code repository as a pull request. The proposal described an overflow in the same AVX-512 Base64 decoder code. Its author closed it without merging it in early November, after an automated notice that it had been inactive for more than eight weeks. Found this article interesting? Follow us on Google News, Twitter and LinkedIn to read more exclusive content we post.
Java4Shell (CVE-2026-114938) Enables Pre-Auth RCE in Jira, Jenkins, Keycloak on Java 17+ AVX-512 VBMI Servers
Full Article
Original Source
Read the full article at Thehackernews →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.