Java4Shell (CVE-2026-114938): Three Vendor Posts Disagree on Whether OpenJDK Has a Fixed Build
Security / news
Java4Shell (CVE-2026-114938): Three Vendor Posts Disagree on Whether OpenJDK Has a Fixed Build
Calif's Base64 decoder bug reaches code execution on AVX-512 servers from OpenJDK 17.0.6 up; Qualys lists fixes that Semgrep and Sonatype do not.
Researchers at Calif disclosed Java4Shell (CVE-2026-114938) on Nov. 17: a memory-corruption bug in OpenJDK's Base64 decoder that gives a remote attacker code execution on x86-64 servers with AVX-512 VBMI, in every OpenJDK build from 17.0.6 onward that lacks the fix. The sources disagree on whether fixed builds exist yet.
The attacker needs only to get malformed Base64 into code that decodes it. They need no account on the targets Calif tested, and the exploitability conditions are the CPU, the build and whether the application decodes untrusted Base64.
Which JDKs and which processors
Semgrep's write-up (published Nov. 17, by Katie Paxton-Fear, Kurt Boberg and Cris Thomas) says Calif "confirmed an out-of-bounds heap write" in handwritten assembly that speeds up Base64 decoding with AVX-512 instructions. Qualys, in a post by Saeed Abbasi, head of its Threat Research Unit, says a short malformed input writes 62 bytes past the end of a buffer before Java rejects it.
| Condition | Affected | Not affected |
|---|---|---|
| OpenJDK version | 17.0.6 and later, including 21, 25 and 27 | Before 17.0.6, including standard 8 and 11 |
| Processor | x86-64 with AVX-512 VBMI (Intel Ice Lake to Granite Rapids; AMD Zen 4 and newer) | ARM64; x86-64 without AVX-512 VBMI |
Qualys adds one Oracle build outside that range: Oracle Java SE 8 Enterprise Performance Pack 8u511b33-perf.
Two scores, and a fix that sources describe differently
Calif and Semgrep score it 9.2 under CVSS 4.0. Oracle's own rating, per Qualys and Sonatype, is 8.1 under CVSS 3.1, and Sonatype says Oracle classes it as difficult to exploit. The two numbers come from different versions of the scoring system and do not contradict each other.
On patches the three posts do not line up. Semgrep says no fixed release numbers were available when it published. Qualys lists Oracle builds released Nov. 17: 8u513-perf, 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1, plus Amazon Corretto 17.0.21.1.2. Sonatype lists Oracle's November Critical Patch Update builds ending in .0.3 as the affected ones, and says Oracle's pre-release notice does not map the Java fixes to this CVE by name.
Treat a build as fixed only after the vendor's advisory says so for this CVE. Other vendors publish on their own schedules, according to Qualys.
Exploitation: demonstrated, not yet observed
Semgrep reports that Calif demonstrated pre-authentication exploitation in default configurations of Atlassian Confluence and Jira, JFrog Artifactory, Jenkins, Keycloak and PingFederate, and post-authentication exploitation against Apache Kafka Broker, Elasticsearch and Gerrit. Qualys adds Solr and Spinnaker to the authenticated list.
Neither post reports exploitation in the wild. Qualys says there was no public confirmation as of Nov. 17, and that the researchers are withholding product-specific exploits until organisations have had time to upgrade. Qualys also says CISA issued an alert urging immediate patching; The Terminal did not fetch that alert.
What to do before a patch is confirmed
Patch first, then restart: a running JVM keeps the old runtime, and container images need rebuilding on a patched base. Until then there are two JVM flags, each needing a restart.
-XX:-UseBASE64Intrinsicsdisables the fast decoder. Qualys and Sonatype prefer it, and it may slow applications that decode a lot of Base64.-XX:UseAVX=2caps AVX use. It works only on Intel and AMD, and a JVM on ARM may refuse to start if the flag is shared across a mixed fleet.
Semgrep adds two indicators to review, which are not proof of compromise: unexplained hs_err_pid crash files, and bursts of IllegalArgumentException from java.util.Base64$Decoder. A reverse proxy is not a control, according to Qualys.
The number to watch is the fixed-build list in Oracle's, Amazon's and each other vendor's advisory, since the three posts above disagree on it. For other recent flaws in widely deployed runtimes and servers, see Ruby 4.0.10's seven regexp and string flaws and nginx2root.
Sources
More in Security
- 01Ruby2Shell (CVE-2026-132825): Ruby Is Fixed, but GitLab, GitHub Enterprise and Chatwoot Are Listed Without a PatchA one-byte heap overflow in Ruby's case conversion carries a 7.4 local score, while the researchers say stock Rails is remotely reachable and will publish a proof of concept Nov. 25.
- 02NPCI Wants Overlay Permissions Restricted on Sideloaded Apps; Google's 2024 India Block List Omits ThemThe reported ask covers accessibility, device admin and display-over-apps. The pilot Google announced on Oct. 3, 2024 named four other permissions.
- 03CVE-2026-99312: A KVM Nested-AMD Bug Scores 8.8, With Fixes Listed for Three Kernel BranchesThe Linux CNA's scoring notes describe a freed page-table root that a guest can aim at; the commit message describes only a stale MMU context.
- 04CISA's Nov. 18 KEV Additions Include Apache Axis 1.x, Which Has No Fix, With a Nov. 21 DeadlineThree of four new entries carry a three-day federal deadline; LiteLLM's advisory names two different patched versions.