Server Online
Server Online

Advisory 2026-11/04 OpenJDK Java4Shell Remote Code Execution (CVE-2026-114938)

Vulnerability Type

Heap memory corruption in the OpenJDK Base64 decoder, known as "Java4Shell" (CVE-2026-114938), in the Java runtime used by FileCloud Content Search (Solr).

Severity factors

OpenJDK rates CVE-2026-114938 high (CVSS 3.1 8.1 / CVSS 4.0 9.2). In FileCloud, it can be triggered by a user who uploads a crafted file while Content Search is enabled. The most likely result is a Content Search crash. Remote code execution cannot be ruled out. It only affects servers whose CPU supports AVX-512 VBMI. The Document Converter (document preview) is not affected. No exploitation was known as of .

Versions affected

All FileCloud Server versions up to and including 23.263.3, on Windows and Linux, with Content Search enabled and running on Java 17.0.6 or later without the November 2026 security update.

Version fixed

Windows: to be released (FileCloud 23.264, due December 2026, bundling a patched Java 25).

Linux: available now through your operating system's OpenJDK security update.

Description

On , OpenJDK and Oracle released fixes for CVE-2026-114938, known as "Java4Shell". It is a memory corruption flaw in Java's Base64 decoder, in an optimized code path that Java uses only on Intel or AMD x86-64 processors that support AVX-512 VBMI. These include Intel Ice Lake and AMD Zen 4 and later, which covers most current servers and cloud VMs. The flaw affects Java 17.0.6 and later, 21, 25 and 27.

FileCloud Server uses Java for Content Search (Solr) and the Document Converter. On Windows, FileCloud bundles OpenJDK 26.0.1 in C:\xampp\java. On Linux, both services use the Java installed from the operating system. In both cases, vulnerability scanners will report the Java runtime as affected.

Whether a FileCloud server is exposed depends on how it is set up:

  • Content Search enabled: affected. Uploaded files are parsed by Content Search. Some document types, such as password-protected Office files, can carry data that reaches the vulnerable decoder. Any user who can upload a file, including through a public upload link, can trigger the flaw.

  • Content Search not enabled: not affected.

  • Document Converter (document preview): not affected. It never passes uploaded content to the vulnerable decoder.

  • Servers without AVX-512 VBMI, or ARM64 servers: not affected, because Java does not use the vulnerable code path on them.

Content Search accepts connections only from the FileCloud server itself, so it cannot be attacked directly over the network. The risk comes from files that users upload.

Interim mitigation

Windows

Interim mitigation for Windows

Content Search (Solr)

  1. Check the Java version FileCloud uses, and whether Content Search is running:

    C:\xampp\java\bin\java -version
    netstat -an | findstr :8983
    

    Java 17.0.6 or later without the November 2026 security update is affected, including the bundled 26.0.1. If nothing is listening on port 8983, Content Search is not running on this server; skip to the Document Converter steps.

  2. Open Notepad as administrator, open C:\xampp\solr\bin\solr.in.cmd, and save a backup copy. If FileCloud is installed in a different folder, use <install folder>\solr\bin\solr.in.cmd.

  3. Add this line at the end of the file, then save it:

    set SOLR_OPTS=%SOLR_OPTS% -XX:-UseBASE64Intrinsics
    
  4. Restart Content Search:

    1. In the FileCloud Control Panel, click Stop next to Content Search.

    2. In Task Manager, on the Details tab, end any java.exe process that is still running Solr. To see which process is Solr, add the Command line column.

    3. In the Control Panel, click Start next to Content Search.

  5. On the FileCloud server, open http://localhost:8983/solr/. On the Dashboard page, confirm that -XX:-UseBASE64Intrinsics is listed under JVM → Args. Then run a search in FileCloud to confirm that Content Search still works.

Note: Restarting Content Search from the Control Panel does not always stop the old Java process. If the setting does not appear on the Dashboard, make sure no Solr java.exe process was left running, then start Content Search again.

Document Converter (optional)

The Document Converter is not affected. You can apply the same setting as an extra precaution.

  1. Open a Command Prompt as administrator and add the setting to the Document Converter service:

    reg add HKLM\SYSTEM\CurrentControlSet\Services\fcdocservice /v Environment /t REG_MULTI_SZ /d "JAVA_TOOL_OPTIONS=-XX:-UseBASE64Intrinsics" /f
    

    This sets the option for the Document Converter service only. It does not affect other Java programs on the server, and it stays in place when the service is restarted from the FileCloud Control Panel.

  2. Restart the Document Converter:

    net stop fcdocservice
    net start fcdocservice
    

    You can also stop and start Document Preview from the FileCloud Control Panel.

  3. To confirm, open PowerShell as administrator and run:

    $p = Get-CimInstance Win32_Process -Filter "Name='java.exe'" | Where-Object CommandLine -match 'FCDocConverter'
    & C:\xampp\java\bin\jcmd.exe $p.ProcessId VM.flags -all | Select-String 'UseBASE64Intrinsics'
    

    The setting is active if the output shows UseBASE64Intrinsics = false. Then open a document preview in FileCloud to confirm previews still work.

Note: To remove the setting later, run reg delete HKLM\SYSTEM\CurrentControlSet\Services\fcdocservice /v Environment /f and restart the Document Converter.

Linux (RHEL 9 and Ubuntu)

Interim mitigation for Linux

Step 1: Check the server

ss -ltn | grep 8983
java -version

If nothing is listening on port 8983, Content Search is not running on this server. You should still update Java.

Step 2: Update Java

RHEL 9

  1. Check what is installed now:

    rpm -qa 'java-17-openjdk*'
    
  2. Upgrade only the Java 17 packages:

    sudo dnf update --disablerepo='filecloud-*' 'java-17-openjdk*'
    

    The FileCloud repository pins the Java version, so without --disablerepo this command reports "Nothing to do". Disabling it applies to this command only; the next dnf update uses the FileCloud repository again.

  3. Confirm that the fix is installed:

    rpm -q --changelog java-17-openjdk-headless | grep -m1 CVE-2026-114938
    

    If the command prints the CVE number, the fix is installed. If it prints nothing, Red Hat may not have published the update for RHEL 9 yet; check the Red Hat CVE page for CVE-2026-114938 and apply the interim mitigation (Step 4) until it is available.

Ubuntu 22.04

  1. Check what is installed now:

    Bash
    dpkg -l 'openjdk-17*' | awk '/^ii/{print $2, $3}'
    
  2. Refresh the package lists, then upgrade only the installed Java 17 packages:

    Bash
    sudo apt update
    sudo apt install --only-upgrade $(dpkg -l 'openjdk-17*' | awk '/^ii/{print $2}')
    

    --only-upgrade upgrades the Java 17 packages you already have and never installs new ones.

  3. Confirm that the fix is installed:

    apt changelog openjdk-17-jre-headless 2>/dev/null | grep -m1 CVE-2026-114938
    

    If the command prints the CVE number, the fix is installed. If it prints nothing, Ubuntu may not have published the fix for Ubuntu 22.04 yet. Check with apt-cache policy openjdk-17-jre-headless and the Ubuntu CVE page for CVE-2026-114938, and apply the interim mitigation (Step 4) until it is available.

Step 3: Restart Content Search and the Document Converter

Programs that are already running keep using the old Java until they restart.

  1. List the running Java processes and when they started:

    ps -eo pid,lstart,cmd | grep '[j]ava'
    
  2. Restart Content Search and the Document Converter:

    sudo systemctl restart solr.service
    sudo systemctl restart fcdocconverter.service
    sudo systemctl status solr.service fcdocconverter.service
    
  3. Check that no other service is still using the old Java:

    # RHEL
    sudo dnf needs-restarting -s
    
    # Ubuntu
    sudo needrestart -r l
    

    Restart any other Java service that is listed. Avoid restarting every listed service at once on a production server, because the list can include Apache or MongoDB.

  4. Run java -version to confirm the new version, then run a search and open a document preview in FileCloud to confirm that both still work. No further steps are needed.

Step 4: Only if you cannot install the update

Content Search (Solr)

Bash
sudo cp /etc/default/solr.in.sh /etc/default/solr.in.sh.bak
echo 'SOLR_OPTS="$SOLR_OPTS -XX:-UseBASE64Intrinsics"' | sudo tee -a /etc/default/solr.in.sh
sudo systemctl restart solr.service
ps -ef | grep [s]olr | grep -o -- '-XX:-UseBASE64Intrinsics'

The last command prints -XX:-UseBASE64Intrinsics if the option is active. Then run a search in FileCloud to confirm that Content Search still works. If Solr does not start, check the line you added for typos, or restore the backup with sudo cp /etc/default/solr.in.sh.bak /etc/default/solr.in.sh and restart Solr.

Document Converter (optional)

The Document Converter is not affected. You can apply the same setting as an extra precaution.

  1. Back up the service file:

    sudo cp /etc/systemd/system/fcdocconverter.service /root/fcdocconverter.service.bak
    
  2. Open /etc/systemd/system/fcdocconverter.service for editing.

  3. Find the ExecStart line and add -XX:-UseBASE64Intrinsics right after java:

    ExecStart=/bin/sh -c '/usr/bin/env HOME=/var/www/html java -XX:-UseBASE64Intrinsics -Djava.library.path="/usr/lib/libreoffice/program/" -jar /opt/app/fcdocconverter/FCDocConverter.jar'
    

    Keep anything else already on the line, such as a port number at the end.

  4. Save the file, then run:

    sudo systemctl daemon-reload
    sudo systemctl restart fcdocconverter.service
    sudo systemctl status fcdocconverter.service
    
  5. To confirm, run:

    Bash
    ps -ef | grep [F]CDocConverter | grep -o -- '-XX:-UseBASE64Intrinsics'
    

    If the setting is active, the command prints -XX:-UseBASE64Intrinsics. Then open a document preview in FileCloud to confirm previews still work.

Deployment notes for interim mitigation

Single-server installations: Apply the steps once, if Content Search is enabled.

Content Search on a separate server, or Solr high availability: Apply the steps on every Solr server. To avoid a search outage, change one server at a time, and confirm that search still works before moving on.

Multitenant deployments: All tenants share the same Content Search, so the steps only need to be applied once on each Solr server.

If you cannot apply the mitigation: Stop Content Search until you upgrade. Search will not return results while it is stopped. Limiting public upload links reduces exposure but does not remove it, because any user who can upload a file can trigger the issue.

FileCloud Online: FileCloud has applied the fix to all FileCloud Online environments. No action is required.

Fix

  • FileCloud Server on Windows, up to and including 23.263.3: We recommend upgrading to 23.264 when it is released (due December 2026). It bundles a patched Java 25. After upgrading, you can remove the line you added to solr.in.cmd, or leave it in place.

  • FileCloud Server on Linux: Install your operating system's OpenJDK security update now, then restart Content Search and the Document Converter, as described in the Linux steps above. On RHEL, use --disablerepo='filecloud-*' for the Java update, because the FileCloud repository pins the Java version.

  • Java that you installed and manage yourself (for example, from jdk.java.net/archive): Replace it now with a patched build from a vendor that provides security updates, such as Amazon Corretto 25.0.5.1.2 or later. Do not switch to an older Java version to avoid this issue.

  • FileCloud Online: FileCloud has applied the fix to all FileCloud Online environments. No action is required.

If you have any questions about this advisory, please contact FileCloud support.