Welcome to our

Cyber Security News Aggregator

.

Cyber Tzar

provide a

cyber security risk management

platform; including automated penetration tests and risk assesments culminating in a "cyber risk score" out of 1,000, just like a credit score.

Java4Shell (CVE-2026-114938): What Is Confirmed, Who Is Exposed, and How to Hunt It

published on 2026-11-17 23:11:32 UTC by Mayuresh
Content:

Java4Shell (CVE-2026-114938) is a heap out-of-bounds write in the x86_64 HotSpot Base64 decoder that uses AVX-512 VBMI. Oracle fixed it on November 17, 2026 and scored it CVSS 3.1 8.1 (AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H), component Compiler. It affects Oracle Java SE 17, 21, 25 and 27, the 8u511 performance build, and GraalVM. Standard Java 8 and Java 11 are not affected. It is not in CISA KEV (catalog 2026.11.17, released 2026-11-17T17:53:15Z) yet. In this post we see the root cause, read from the OpenJDK fix commit itself, and what it means for exposure. Two findings that matter for defenders – Java 21, 25 and 27 reach the vulnerable code with much shorter input than Java 17. Just over 31 characters instead of over 128 and hosts on Intel Skylake-SP and Cascade Lake, which have AVX-512 but not VBMI are not affected.

Java4Shell (CVE-2026-114938)
Date (UTC)Update
18 Nov 2026Public reproducers are out. At least four GitHub repositories now trigger the heap overflow. One reports a JVM crash in 3 of 4 attempts on 21.0.13.0.3, and no crash with -XX:-UseBASE64Intrinsics. No public repository includes a code execution chain, and Calif is withholding its exploits. Exploitation in the wild is still not reported.
18 Nov 2026More vendors are fixed. Azul Zulu shipped 17.0.21.1+2, 21.0.13.1+2, 25.0.5.1+2 and 27.0.1.1+4 in its November update. Eclipse Temurin released 21.0.13.1+2 and 27.0.1.1+4; Temurin 17 and 25 are still early-access builds only. BellSoft Liberica tagged 17.0.21.1+6, 21.0.13.1+6, 25.0.5.1+5 and 27.0.1.1+5. Red Hat has not changed RHSB-2026-018 and has no fixed packages yet.
18 Nov 2026Amazon Linux is fixed. java-17, java-21, java-25 and java-27-amazon-corretto are fixed on Amazon Linux 2023 (ALAS2023-2026-3478 to 3481) and Amazon Linux 2027 Preview (ALAS2027-2026-371 to 374); java-17-amazon-corretto is fixed on Amazon Linux 2 Core (ALAS2-2026-4123). The java-1.7.0, java-1.8.0 and java-11 Corretto packages are not affected. java-11-openjdk and java-22, 23, 24 and 26-amazon-corretto are marked no fix planned. Amazon scores the flaw CVSS 3.1 8.8 (AV:N/AC:L/PR:L), which differs from Oracle’s 8.1 (AC:H/PR:N).
18 Nov 2026Vendor sweep. The fix is in the source at these release tags: OpenJDK jdk-17.0.21.1-ga, jdk-21.0.13.1-ga, jdk-25.0.5.1-ga and jdk-27.0.1.1-ga; Amazon Corretto 17.0.21.1.2, 21.0.13.1.2, 25.0.5.1.2 and 27.0.1.1.4; SAP SapMachine 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1. Eclipse Temurin has early-access builds only. Azul Zulu, BellSoft Liberica and Ubuntu have no fixed release yet. Red Hat published RHSB-2026-018 with no fixed packages yet.
18 Nov 2026Changed the -XX:UseAVX=2 row to Red Hat’s guidance: it is Red Hat’s interim mitigation on x86-64 hosts, and it stops JVMs on ARM64 from starting. Added Red Hat’s list of exposed CPUs. Redrew all three diagrams in the site font.
17 Nov 2026Added fixed builds and patch numbers from Oracle’s Java SE patch document CPU435, including 8u513-perf, GraalVM for JDK 17.0.21.1 and 21.0.13.1, and GraalVM Enterprise Edition 21.3.20.1. Java 7 (7u523) is not affected.
17 Nov 2026Added the disclosure timeline: 1,402 days in shipped Java 17 builds, and 5 days from the fix commit to the advisory.
17 Nov 2026Scored the public checker ashrafzaryouh/CVE-2026-114938: 51 of 100. It is right on Java 17 and passes affected Java 21, 25 and 27 builds with exit code 0. Added PoC questions to the FAQ.
17 Nov 2026Rewrote the post from Oracle’s advisory and the OpenJDK fix JDK-8393212: root cause, affected builds, the AVX-512 VBMI requirement, request-level detections and the -XX:-UseBASE64Intrinsics workaround. Corrected the “Java 17 only” claim and the 65-byte trigger, which reaches the bug on Java 21 and 25 but not on Java 17.
17 Nov 2026First version, from pre-patch reporting. Oracle had not yet named the CVE.

CVE-2026-114938 (Java4Shell)

  • Unauthenticated heap out-of-bounds write, Oracle CVSS 3.1 8.1, Calif CVSS 4.0 9.2
  • Java 21, 25 and 27 reach it with input over 31 characters
  • October 2026 builds are on Oracle's affected list
  • Needs x86_64 with AVX-512 VBMI and a JIT-compiled decoder
  • Code execution claimed by Calif, not reproduced in public
  • Not in CISA KEV 2026.11.17, no exploitation reported

Patch this week

outcome: Move to Oracle 17.0.21.1, 21.0.13.1, 25.0.5.1 or 27.0.1.1, or your vendor's build with JDK-8393212

outcome: Until then, -XX:-UseBASE64Intrinsics

Confirmed by Oracle and the OpenJDK fix

Java4Shell triage as of November 17, 2026

If you patched Java in October, you are still exposed to Oracle’s November Java SE fixes. Oracle lists 8u511, 11.0.33, 17.0.21, 21.0.13, 25.0.5 and 27.0.1 as affected.

What is confirmed about Java4Shell (CVE-2026-114938), and what is only reported?

The flaw, its scope and its root cause are now confirmed by Oracle and by the OpenJDK fix. Code execution is still only Calif’s claim, and nobody has reported exploitation.

ClaimStatusSource
Calif found and reported Java4ShellConfirmed. Oracle credits Quang Luong of Calif.io, and separately Albert YangOracle advisory
CVE-2026-114938 is one of Oracle’s 11 Java SE fixesConfirmed. CVSS 3.1 8.1, component CompilerOracle advisory
Heap out-of-bounds write in the AVX-512 Base64 decoderConfirmed by the fix, JDK-8393212jdk17u commit 8ea73158af
62-byte overflowReproduced independently on Intel Ice LakeUpwind
CVSS 4.0 score of 9.2Calif’s score; Oracle uses CVSS 3.1 and scores 8.1Calif
Remote code execution and a Heartbleed-style memory readClaimed by Calif and CISA. Upwind reproduced the overflow but not code executionCalif, Upwind, CISA
A 65-byte input triggers itHalf right. A 65-byte input (64 plus a 1-byte tail) reaches the bug on Java 21 and 25, where the minimum is over 31; on Java 17 the minimum is over 128My inference of the fix, below
“Java 17 only”Wrong. Oracle lists 17, 21, 25, 27 and 8u511b33-perfOracle advisory
Exploited in the wildNo. Not in CISA KEV 2026.11.17, no reportsCISA KEV
Public proof of conceptNone found on November 17Searches on GitHub and the web

The name “Java2Shell” also circulates. No primary source uses it. Calif’s name for the bug is Java4Shell.

Before the patch, several write-ups stated “Java 17 only” as fact, and it was wrong. The reported 65-byte trigger was half right: it works on Java 21 and 25, not on Java 17. Oracle has not assigned a CWE yet. My inference from the fix, is CWE-191 (integer underflow) leading to CWE-787 (out-of-bounds write).

Which Java versions and platforms are affected by Java4Shell?

Oracle lists Java SE 17, 21, 25 and 27, the 8u511 performance build, and GraalVM as affected by CVE-2026-114938. Standard Java 8 and Java 11 are not on the list, because they do not have the AVX-512 Base64 decoder.

Product rangeAffected software Fixed build (Oracle patch number)Minimum input length to reach the bug
Java SE 2727.0.1.0.327.0.1.1 (Patch 40239622)Not checked; same code lineage as 25
Java SE 2525.0.5.0.325.0.5.1 (Patch 40239624)Over 31 characters
Java SE 2121.0.13.0.321.0.13.1 (Patch 40239626)Over 31 characters
Java SE 1717.0.21.0.3 (code present since 17.0.6)17.0.21.1 (Patch 40239631)Over 128 characters
Java SE 8 Enterprise Performance Pack8u511b33-perf8u513-perf (Patch 40239685)Not checked
Oracle GraalVM for JDK17.0.21.0.3, 21.0.13.0.317.0.21.1, 21.0.13.1Not checked
Oracle GraalVM Enterprise Edition21.3.20.0.321.3.20.1Not checked
Java SE 11Not affected11.0.33.1 fixes other November CVEsNo AVX-512 decoder
Java SE 8Not affected8u513 fixes other November CVEsNo AVX-512 decoder
Java SE 7 (Market Driven Support)Not affected7u523 fixes CVE-2026-114939, CVE-2026-114944 and CVE-2026-114946 onlyNot applicable

The affected builds come from the Oracle advisory’s Java SE risk matrix. The fixed builds and patch numbers come from Oracle’s Java SE patch availability document, CPU435, which needs a My Oracle Support login. I confirmed the OpenJDK side myself: the jdk17u tag jdk-17.0.21.1-ga contains the fix. The minimum lengths are from my reading of the jdk17u, jdk21u and jdk25u source. The CPU decides exposure as much as the version does. The vulnerable code runs only on x86_64 CPUs with AVX-512 VBMI: Intel Ice Lake server and later, and AMD Zen 4 and later. Intel Skylake-SP and Cascade Lake have AVX-512 but no VBMI, so they skip it. AVX2-only machines and ARM servers such as AWS Graviton are not affected; the fix touches only the x86 stub.

OpenJDK distributions from Temurin, Corretto, Zulu, Microsoft and Red Hat ship on their own schedules. Look for JDK-8393212 in your vendor’s release notes rather than matching Oracle’s version numbers.

What is the root cause of Java4Shell?

The AVX-512 VBMI Base64 decoder in HotSpot computes a negative output size when a malformed input ends with a padded tail of 1 to 3 characters. The negative size turns its 64-byte write mask into all ones, so the decoder stores 64 bytes past the point where it should store almost nothing. The OpenJDK fix is just four lines long. The fix is JDK-8393212, “Base64 optimization improvements”, committed to jdk17u on November 12, 2026 as 8ea73158af, a backport of mainline commit e56f69c725f8. It adds one check at the start of the decoder’s final-fragment path in stubGenerator_x86_64.cpp:

__ BIND(L_finalBit);
// Now have 1 to 63 bytes left to decode

// A padded Base64 tail shorter than one complete block is handled by Java.
__ cmpl(length, 4);
__ jcc(Assembler::less, L_bruteForce);

Here is what happens without that check, from my reading of the same file:

  • bzhiq builds the store mask from the output size. With -1 or -2 as the bit index, it clears nothing, so all 64 mask bits stay set. The masked store evmovdqub then writes all 64 bytes of the vector register into the destination array.ver enters the vector loop that write-ups describe as the bug. It would take the stub’s scalar loop, or a padding path in the Java caller. Either the trigger size is wrong, or the bug is not in the vector code. I cannot tell which until the fix commit is public.
  • The vector path runs only on CPUs with AVX-512 VBMI, for non-MIME input longer than 128 bytes on Java 17. On Java 21 and 25 the same entry check is cmpl(length, 31), so input longer than 31 bytes is enough.
  • The decoder consumes the input in 64-byte blocks. Whatever is left, 1 to 63 bytes, goes to L_finalBit.
  • It sets the output size to (length / 4) * 3. For a tail of 1 to 3 bytes that is 0.
  • If the last byte is =, the L_padding code subtracts 1 from the output size, and 1 more if the byte before it is also =. The size becomes -1 or -2.

The Java caller passes the raw input length to the intrinsic, not a length rounded to a multiple of 4, so a malformed tail reaches the stub. A valid Base64 never has a tail of 1 to 3 characters, which is why Upwind saw no overflow from valid input in 300,000 runs. Intel’s AVX-512 decoder, JDK-8268276, was committed in mainline JDK 18 and was backported to 17.0.6 build 03 (JDK-8268276, backport review). A separate data-loss bug in the same stub, JDK-8321599, was fixed in January 2024 (integration notice). Java4Shell sat in this path for about four years on Java 17.

How long was Java4Shell in Java 17?

Java4Shell shipped in Java 17 builds for 1,402 days, from the 17.0.6 GA tag on January 10, 2023 to the fix commit on November 12, 2026. The fix then stayed private for 5 days. Calif’s leak statement, Oracle’s advisory and the public fix all landed on November 17, 2026.

Java4Shell timeline in days

  • Backport commit to fix commit: 1464Nov 9, 2022 to Nov 12, 2026
  • 17.0.6 GA tag to fix commit: 1402Jan 10, 2023 to Nov 12, 2026
  • Missed bug fix in the same stub to fix commit: 1039JDK-8321599, Jan 8, 2024
  • Fix commit to public advisory: 5Nov 12 to Nov 17, 2026
  • Calif leak statement to public fix: 0both Nov 17, 2026
Source: OpenJDK jdk17u and jdk17u-dev git history, Oracle CSPU November 2026
DateEventSource
November 17, 2026Upwind reproduces the 62-byte overflow; CVE-2026-114938 not in CISA KEV 2026.11.17Upwind, KEV file
November 17, 2026Fix made public with the jdk-17.0.21.1-ga tag; earlier that day the public jdk-17.0.21.1+1 tag did not contain itMy checks of openjdk/jdk17u
November 17, 2026Calif says the bug leaked; Oracle’s advisory names CVE-2026-114938Calif on X, Oracle
November 16, 2026South Korea NCSC advisory, reported second-hand and unverifiedEndor Labs
November 12, 2026Fix JDK-8393212 committed to jdk17u (8ea73158af)Commit
Not publicCalif reports the bug to OracleOracle credits Calif.io and Albert Yang; no date given
January 8, 2024Data-loss bug JDK-8321599 fixed in the same stub; Java4Shell missedjdk17u-dev commit 1f95796210
January 10, 202317.0.6 GA tagged: the bug shipsjdk17u-dev tag jdk-17.0.6-ga
November 15, 2022First Java 17 build with the decoder: 17.0.6+3jdk17u-dev tag jdk-17.0.6+3
November 9, 2022AVX-512 Base64 decoder backported to jdk17u (JDK-8268276)jdk17u-dev commit 07050b000c

This is how the timeline looks:

How would a Java4Shell attack work?

An attacker sends malformed Base64 to a Java service until the JIT compiles the decoder, then triggers a 62-byte heap overflow. Code execution is Calif’s claim alone.

  1. Find a target: Any Java service that decodes client-supplied Base64 with java.util.Base64: Authorization headers, tokens, cookies, API fields. Enterprise applications such as Jenkins, Jira and Keycloak need to be checked for.
  2. Warm up the decoder. The vulnerable code is a JIT intrinsic, so the decoder must be compiled first. A busy authentication endpoint is already warm.
  3. Send the trigger. Malformed Base64 over 128 characters on Java 17, or over 31 on Java 21 and 25, ending in = with a length that is not a multiple of 4.
  4. Corrupt the heap. The decoder writes 64 bytes where it should write almost nothing.
  5. Reach code execution. Calif says the overflow leads to remote code execution and a Heartbleed-style memory read.
Java4Shell (CVE-2026-114938) attack path.

Which hosts can reach the Java4Shell code path?

A host is exposed only if the following conditions are met and each one has an exit that stops the attack. The CPU gate removes many older servers outright, and the JVM flag gate is a mitigation you control today.

ConditionQuestionExit that stops the attackBasis
1Does attacker data reach the java.util.Base64 basic or URL decoder?No attacker Base64, or only the MIME decoder, which takes the scalar pathStub entry checks isMIME
2Is it an x86_64 CPU with AVX-512 VBMI?Skylake-SP, Cascade Lake, AVX2-only x86 and aarch64 never run the vector pathVector code sits inside the VBMI branch
3Is UseBASE64Intrinsics on? It defaults on with UseAVX > 2 and AVX-512 VL plus BW-XX:-UseBASE64Intrinsics or -XX:UseAVX=2vm_version_x86.cpp;
4Has the JIT compiled the decoder?Cold code runs the Java implementationNeeds about 10,000 decodes
5Is the input over 128 characters (Java 17) or over 31 (Java 21, 25), with a 1 to 3 character tail ending in =?Short input or valid Base64Stub entry check and the fix
6Is the build missing JDK-8393212?Patched buildOracle risk matrix, jdk17u tag jdk-17.0.21.1-ga

A login or token endpoint decodes Base64 on every request, so it is compiled soon after startup, and an attacker can warm up a quiet endpoint with ordinary requests. Condition 5 is where Java 21 and later differ most from Java 17: a malformed 33-character token in a Basic authentication header already clears it.

Java4Shell (CVE-2026-114938) reachability conditions

How do I check if a host is exposed to Java4Shell?

Check four things on each host: the Java build, whether the CPU has AVX-512 VBMI, whether UseBASE64Intrinsics is on, and whether the trigger shape already reached the application. Every command below only reads. Run the hunting commands before you patch, because a restart clears the evidence.

Linux

# 1. Every JDK on the host, including bundled ones (reads each JDK's release file)
find / -xdev -name release -path '*j*' 2>/dev/null | xargs grep -H '^JAVA_VERSION=' 2>/dev/null

# 2. Does the CPU have AVX-512 VBMI? No avx512vbmi means the vulnerable path cannot run
grep -o -w -E 'avx512vl|avx512bw|avx512vbmi' /proc/cpuinfo | sort -u

# 3. Is the intrinsic on for a given JDK?
/path/to/java -XX:+PrintFlagsFinal -version 2>/dev/null | grep -E 'UseBASE64Intrinsics|UseAVX '

# 4. Same check against a running JVM
jcmd <pid> VM.flags -all | grep UseBASE64Intrinsics

# 5. Did the trigger shape reach the application? (adjust the log path)
grep -R -n 'Input byte array has wrong 4-byte ending unit' /var/log /opt/*/logs 2>/dev/null | head

# 6. Recent JVM crash logs
find / -xdev -name 'hs_err_pid*.log' -newermt 2026-11-01 2>/dev/null

# 7. Child processes of Java
ps -eo pid,ppid,user,lstart,cmd --forest | grep -B1 -A3 '[j]ava'

What detections work for Java4Shell?

The fix tells us the exact input shape, so request-level detection is now possible. A Base64 token that ends in = but whose length is not a multiple of 4 is always invalid, and it is the shape that reaches the bug. I tested the pattern below against 7,980 valid Base64 strings with no matches, and it matched all 1,656 malformed samples I generated.

ModSecurity (java4shell_modsecurity.conf). Regex tested in Python; not yet run on a ModSecurity engine.

SecRule REQUEST_HEADERS:Authorization "@rx (?i)^(?:basic|bearer)\s+(?:[A-Za-z0-9+/_-]{4}){8,}[A-Za-z0-9+/_=-]{0,2}=$" \
    "id:1146938,phase:1,log,pass,t:none,\
    msg:'Possible Java4Shell (CVE-2026-114938) malformed Base64 in Authorization header',\
    tag:'CVE-2026-114938',severity:'CRITICAL'"
SecRule ARGS|REQUEST_COOKIES "@rx ^(?:[A-Za-z0-9+/_-]{4}){8,}[A-Za-z0-9+/_=-]{0,2}=$" \
    "id:1146939,phase:2,log,pass,t:none,t:urlDecodeUni,\
    msg:'Possible Java4Shell (CVE-2026-114938) malformed Base64 in parameter or cookie',\
    tag:'CVE-2026-114938',severity:'CRITICAL'"

The rules log and pass. Switch pass to deny once you have watched them for a day; valid Base64 never has this shape, so the only other hits should be buggy clients.

Java4Shell GitHub repositories: which ones are safe and which ones work

Within a day of disclosure, 40 public GitHub repositories carried Java4Shell code, checkers or rules. I ran my repository reliability score over all of them, the same method I used for PHP2Shell. Every repository gets a safety status. Defensive repositories also get a score out of 100. Exploit repositories get a safety status only, and I do not rank them.

Snapshot: the nomi-sec PoC-in-GitHub index as of Nov 18, 2026 13:56 UTC (38 repositories naming CVE-2026-114938) plus two repositories named only “Java4Shell”. GitHub search was not available to me for this run, so repositories named after the nickname alone are undercounted. A second pass runs on Nov 25, 2026. CVE-2026-114938 is not in the CISA KEV catalog (catalogVersion 2026.11.17, released 2026-11-17T17:53:15Z).

Information about the 40 repositories

Java4Shell repositories by class

  • dark: Local version checkers = 13 dark: Detection rules: 6one is unscored, see below dark: Mitigations = 4 | WAF, RASP agent, runtime patch, build flag guard grey: Documentation and trackers = 2 grey: Source patch = 1 grey: Other = 1 group: red: Proof-of-concept labs = 8 red: Exploit repositories = 3 | one named as remote code execution red: Crash reproducers = 2

Classes from file names, metadata and defensive README files. Exploit-side classes come from metadata only.

40 repositories, nomi-sec snapshot of Nov 18, 2026 plus two nickname-only repositories

Classification: A scanner of the installed-package kind: one Python file that finds Java binaries, runs java -version on each and grades the version string. It has no network code and no exploit code. Its README is preliminary and still says the root cause is unconfirmed.

One thing stands out against PHP2Shell. Almost every defensive tool here is a local version checker, not a remote probe. That is the right call for this bug. Nothing on the wire tells you whether the JVM on the other end runs the AVX-512 VBMI intrinsic, so a remote banner check can only ever say “this version, maybe”.

Safety status

33 repositories came back clear, which means I found no red flags, not that they are proven safe. Seven get a caution:

RepositoryClassCaution reasons
theori-io/java4shell-runtime-patch-CVE-2026-114938MitigationREADME installs with a download piped into sh from files[.]xint[.]io; prebuilt jar in the repository. The script pins hashes, which helps, but you still run what the server sends.
jeffschafer/MANUAL-Scanner-Java4Shell-CVE-2026-114938CheckerThree prebuilt zip archives. I listed them and checked SHA256SUMS: they hold the same scripts as the repository and the checksums match.
WaveShift/ZenithJvmFlagGuardMitigationPrebuilt gradle-wrapper.jar that I could not check against gradle.org from my sandbox.
jiparis/java4shellOtherCI workflow pipes an installer download into bash.
darkfibr/yara-java4shellDetectionLabelled as a YARA rule but also ships a trigger program and a binary capture. Unscored until I review it by hand.
zejunyang/java4shell-rceExploitPrebuilt jar in dist/.
Pwned-test/cve-2026-114938-java17-PoCProof-of-concept labPrebuilt .class files in bin/.

No repository decodes and runs an embedded payload, none promotes a Telegram or Discord group, and none has a commit history staged to look older than the repository. Two curl-into-shell hits, in ionsec/java4shell-detector and CypherSecurity/java4shell-detection, are detection rules and threat notes quoting the pattern, so they stay clear.

I ran the checkers against 12 Java builds

I ran eight local checkers in a sandbox with no network, each as its own unprivileged user, against 11 fake Java homes that report a chosen version and one real OpenJDK 21.0.12.1 on a CPU with AVX-512 VBMI. The targets cover the edges that matter: Java 11 and standard Java 8 (not affected), 17.0.5 (before the bug), 17.0.6 (first affected), the vulnerable 17, 21, 25 and 27 builds, and fixed 17.0.21.1 and 21.0.13.1 builds from Oracle and Temurin.

Right verdicts out of 12 Java builds

  • max=12 green: +bold Mahgal/java4shell-check = 12 green: +bold Ademking/java4shell (host checker) = 12 amber: XanDuri/java4shell-check: 11fixed Temurin build reported vulnerable amber: avoliomattia/CVE-2026-114938 = 9 | fixed builds below +4 reported vulnerable amber: CypherSecurity/java4shell-detection (bundled checker) = 9 | misses the real vulnerable JDK amber: afe038/audit-java4shell = 8 | flags standard Java 8; fixed builds left as review amber: ashrafzaryouh/CVE-2026-114938 = 5 | Java 17 logic only red: ubitquity (bundled check script) = 0 | marks every running process exposed

Fake Java homes report a version only; one target is a real vulnerable JDK on an AVX-512 VBMI CPU. Review counts as not right.

Lab run, Nov 18, 2026

A .1 version number alone does not prove the fix

  • OpenJDK tags jdk-17.0.21.1+0 and +1 do not contain the fix high: jdk-21.0.13.1+0 does not contain the fix mid: First fixed tags: 17.0.21.1+2, 21.0.13.1+2, 25.0.5.1+2, 27.0.1.1+4 low: Vendor GA builds match these (Corretto, Temurin, Azul) outcome: Pre-release and early-access builds can say 17.0.21.1 and still be vulnerable. A checker that compares only the four-part version passes them. Check the +N build number against the first fixed tag.

Check the build number

Certain

One checker calls a vulnerable JDK patched

  • CypherSecurity checker treats any fourth version digit as "patched likely" high: Real OpenJDK 21.0.12.1 on a VBMI CPU came back PATCHED_LIKELY mid: The same logic would pass Oracle 17.0.21.0.3, an affected build low: Its Suricata and Sigma rules load cleanly outcome: Use the rules, not the bundled checker, until the version logic changes.

False negative

Certain

Two lab findings worth acting on

Rules: all Suricata rules in four repositories load in Suricata 7.0.3 (18 rules, zero failures), and all nine Sigma rules in two repositories parse. The darkfibr YARA file compiles. I could not build the Nextron THOR plugin or the Go and TypeScript checkers, because the sandbox blocks module downloads, so those keep static scores.

Reliability scores, by class

Score = 50 content + 30 provenance + 20 hygiene, capped so provenance and hygiene cannot carry a repository whose content fails. Scores only compare within a class. A documentation repository earns points for stating facts, a checker for passing a test. Organization verification is the one indicator I did not collect; organizations get 3 of 5 points for now. The coverage column in the CSV shows how many indicators applied to each repository.

ClassRepositoryScoreLabStars
CheckerMahgal/java4shell-check85correct 12/120
CheckerAdemking/java4shell80correct 12/120
Checkerafe038/audit-java4shell73partial 11/120
Checkeravoliomattia/CVE-2026-11493873partial 9/120
Checkerjeffschafer/BigFix-Java4Shell-CVE-2026-114938-Scanner65not run0
Checkernu0l/Java4Shell-LocalScanner65not run0
Checkertobiastillstam/Scan-Java4Shell65not run0
Checkertechdevfun/java4shell61not run0
Checkerashrafzaryouh/CVE-2026-11493861partial 5/126
Checkerjeffschafer/MANUAL-Scanner-Java4Shell-CVE-2026-11493858not run0
CheckerXanDuri/java4shell-check58partial 11/120
Checkerrogersouzak8s/java4shell-check55needs Docker or Kubernetes0
Checkeryora1928/CVE-2026-114938_by_yora192852not run0
Detectionionsec/java4shell-detector88Suricata 8/8, Sigma 6/60
DetectionNextron-Labs/thor-plugin-jvm-crash-cluster87not built0
DetectionCypherSecurity/java4shell-detection85Suricata 2/2, Sigma 3/30
Detectionarturo-b-cmu/CVE-2026-11493870Suricata 3/33
Detectionjsik22/java4shell-detection60Suricata 5/50
Mitigationboundaryx/cloudrasp-java4shell88not run2
Mitigationcm-suzuki-ryo/aws-waf-java4shell-base64-guard85not run0
Mitigationtheori-io/java4shell-runtime-patch-CVE-2026-11493866not run18
MitigationWaveShift/ZenithJvmFlagGuard62not run0
Patchubitquity/cve-2026-114938-java4shell-remediation42check script failed0
Documentationpathcosmos/CVE-2026-11493885n/a0
Documentationklapertart/Java4Shell-CVE-2026-114938-Summary68n/a0
Otherjiparis/java4shell20n/a0

The ubitquity source patch is not the OpenJDK fix. It bails out to a different label and adds an extra check on the last byte. I did not build it. Use the vendor build instead. What the repositories have in common

No public IP addresses appear anywhere in the 40 repositories; every address-shaped string turned out to be a four-part Java version. Six User-Agent strings appear in code. The one worth a rule is java4shell-scanner/1.0: it shows up first in ikow/java4shell-poc (first commit Nov 17, 2026 23:04 UTC) and again, with a project URL appended, in Ademking/java4shell the next morning. A shared name in an exploit repository and a defensive scanner means a match in your logs does not tell you which one hit you. One exploit repository references an out-of-band interaction domain (burpcollaborator[.]net). The full list, with first-commit times, is in the indicators CSV.

User-AgentRepositoriesFirst seen in code (UTC)
java4shell-scanner/1.0ikow/java4shell-poc; Ademking/java4shell (with project URL)Nov 17, 2026 23:04
java4shell-probe/1.0r3dvei1/java4shellNov 18, 2026 03:48
ionsec-java4shell-detector/1.0ionsec/java4shell-detectorNov 18, 2026 08:27
java2shell-public-research-crawlerzake1god/java2shellNov 18, 2026 04:27

Copy tracing found one near-copy: 76 percent of the distinctive code lines in yashbhure009/Java4Shell-writeup- also appear in ikow/java4shell-poc. Everything else is original at the file and line level.

Patch Java4Shell (CVE-2026-114938)

VendorStatusFixed buildsSource
Oracle Java SE and GraalVMFixed17.0.21.1, 21.0.13.1, 25.0.5.1, 27.0.1.1, 8u513-perf; GraalVM for JDK 17.0.21.1 and 21.0.13.1; GraalVM EE 21.3.20.1Oracle CPU435
OpenJDK (upstream)Fixed in sourcejdk-17.0.21.1-ga, jdk-21.0.13.1-ga, jdk-25.0.5.1-ga, jdk-27.0.1.1-gaFix present in each tag
Amazon CorrettoFixed in source17.0.21.1.2, 21.0.13.1.2, 25.0.5.1.2, 27.0.1.1.4Fix present in each tag
Amazon LinuxFixedjava-17, java-21, java-25 and java-27-amazon-corretto (ALAS2023-2026-3478 to 3481, ALAS2027-2026-371 to 374, ALAS2-2026-4123); java-22, 23, 24 and 26-amazon-corretto: no fix plannedALAS, November 17, 2026
SAP SapMachineFixed in source17.0.21.1, 21.0.13.1, 25.0.5.1, 27.0.1.1Fix present in each tag
Eclipse TemurinNot releasedEarly-access builds only (17.0.21.1+1-ea-beta, 21.0.13.1+1-ea-beta, 25.0.5.1+1-ea-beta)Temurin binaries repositories
Red HatNo fix yetNone; Red Hat is expediting fixesRHSB-2026-018
Azul ZuluNo release foundLatest is the October 24, 2026 out-of-cycle updateAzul release notes
BellSoft LibericaNo release foundLatest tags are 25.0.5.0.3+5 and 27.0.1.0.3+5bell-sw/Liberica tags
UbuntuNo fix foundUSN-9014-1 and USN-9015-1 (November 9) ship the pre-fix 17.0.21.0.3 and 21.0.13.0.3USN notices
DebianNo fix yetopenjdk-17, 21, 25 and 27 vulnerable in bookworm, trixie, forky and sid; no fixed versionDebian security tracker
IBM Semeru (OpenJ9)UnknownNo statement foundNot checked

Amazon will not fix java-22, java-23, java-24 or java-26-amazon-corretto. Those releases are out of support, and they most likely carry the same decoder, since it entered mainline in JDK 18. Move those hosts to Java 25 or 27, or run them with -XX:-UseBASE64Intrinsics until you do.

Java4Shell FAQ

What is Java4Shell (CVE-2026-114938)? Java4Shell is a heap out-of-bounds write in the x86_64 HotSpot Base64 decoder that uses AVX-512 VBMI, found and reported by Calif. Oracle fixed it on November 17, 2026 and scored it CVSS 3.1 8.1. A malformed Base64 input makes the decoder write 64 bytes where almost none fit.

Which Java versions are affected by Java4Shell? Oracle lists Java SE 17.0.21.0.3, 21.0.13.0.3, 25.0.5.0.3, 27.0.1.0.3, 8u511b33-perf, GraalVM for JDK 17.0.21.0.3 and 21.0.13.0.3, and GraalVM Enterprise Edition 21.3.20.0.3. Standard Java 8 and Java 11 are not affected. On Java 17 the code has been present since 17.0.6.

Is Java4Shell being exploited? No exploitation has been reported. CVE-2026-114938 is not in CISA KEV catalog version 2026.11.17, and no public proof of concept was found on November 17, 2026.

How do I fix Java4Shell? Upgrade to Oracle Java 17.0.21.1, 21.0.13.1, 25.0.5.1 or 27.0.1.1, the 8u513 Enterprise Performance Pack, Oracle GraalVM for JDK 17.0.21.1 or 21.0.13.1, or GraalVM Enterprise Edition 21.3.20.1, or to an OpenJDK vendor build whose release notes list JDK-8393212, then restart every JVM.

Is there a workaround for Java4Shell? Add the JVM option -XX:-UseBASE64Intrinsics. It turns off the vulnerable AVX-512 decoder so Base64 decoding runs in Java. Upwind tested it clean on every affected version. It is not Oracle guidance, and it slows Base64 decoding.

How do I check if a server is exposed to Java4Shell? Check that the CPU reports avx512vbmi in /proc/cpuinfo, that java -XX:+PrintFlagsFinal shows UseBASE64Intrinsics set to true, and that the Java build is on Oracle’s affected list. Intel Skylake-SP and Cascade Lake servers have no VBMI and are not exposed.

Why is Java 21 easier to reach than Java 17? The vulnerable vector path starts at inputs longer than 31 bytes on Java 21 and 25, against longer than 128 bytes on Java 17. A short malformed token in an Authorization header is enough on the newer releases.

Is there a public Java4Shell PoC or exploit on GitHub? In the searches I ran on November 17, 2026, I did not find a public Java4Shell trigger or exploit. The repositories I found are defensive: version checkers such as ashrafzaryouh/CVE-2026-114938, a BigFix scanner, and a runtime patch from Xint. Read any script before you run it.

Is the ashrafzaryouh CVE-2026-114938 checker reliable? It is safe to run and correct for Java 17, but it marks affected Java 21, 25 and 27 builds as REVIEW and exits with code 0, and it does not check for AVX-512 VBMI. It scores 51 out of 100 on my PoC repository reliability score.

Article: Java4Shell (CVE-2026-114938): What Is Confirmed, Who Is Exposed, and How to Hunt It - published 4 days ago.

https://pentestit.com/java4shell-cve-2026-114938-detection-mitigation/   
Published: 2026 11 17 23:11:32
Received: 2026 11 20 07:48:29
Feed: PenTestIT
Source: PenTestIT
Category: News
Topic: Security Tooling
Views: 9

Custom HTML Block

Click to Open Code Editor