Purpose of the Template
The flaw in the OpenJDK x86-64 Base64 decoder, tracked as CVE-2026-114938 and known as Java4Shell, highlights that a clean software inventory doesn't guarantee a secure environment. This vulnerability affects only certain x86-64 processors with AVX-512 VBMI instructions. The same application, version, and container image can be vulnerable on one node but safe on another.
This worksheet provides a structured approach to assess exposure per workload. It identifies four critical conditions: an affected runtime, relevant CPU capability, the vulnerable execution path, and attacker-controlled input reaching the decoder. Use this to transform a vague "we run Java" into a prioritized, evidence-backed remediation list. It's applicable to any runtime-level flaw where hardware or distribution context alters exploitability.
Prerequisites
- Access to host and container telemetry, or a sysadmin to run commands.
- A list of Java-based identity and software-delivery systems.
- Vendor contacts for any commercial product embedding a runtime.
- Agreement on a restart or rebuild window, as a patched runtime on disk doesn't fix a running process.
The Template
Copy one block per workload. Treat "unknown" as an answer, not a blank.
JAVA RUNTIME EXPOSURE RECORD
1. WORKLOAD
Name / owner:
Business function:
Holds high-value secrets? (tokens, credentials, keys, session data) Y/N/Unknown
Priority tier (see customization): 1 / 2 / 3
2. RUNTIME IDENTITY
Discovery source: [host scan | container image | vendor product | app directory | other]
Distribution/vendor: [Oracle | Amazon | Eclipse | Microsoft | Red Hat | Azul | embedded in vendor product | unknown]
Exact version string (from `java -version`):
Runtime path on disk or image layer:
Who controls updates to this runtime? [us | product vendor | image maintainer]
3. HARDWARE CONTEXT
Processor model(s) this workload can run on:
AVX-512 VBMI present? [yes | no | varies by node | unknown]
Can the scheduler move this workload between node types? Y/N
Evidence: (command output, node inventory reference)
4. EXECUTION PATH
Does the workload decode Base64 from external data? Y/N/Unknown
Input sources: [inbound requests | file uploads | auth tokens | outbound TLS responses | other]
Makes outbound connections to systems we don't control? Y/N
Outbound destinations: (integrations, connectors, crawlers, batch jobs)
5. DETERMINATION
Exposure status: [Exposed | Likely exposed | Not exposed | Unknown]
Basis for status (one line):
6. REMEDIATION
Correct vendor fix identified: (vendor + fixed version)
Fix owner / ticket:
Method: [vendor patch | image rebuild | product upgrade | runtime swap]
Restart or redeploy completed? (date, evidence)
Verified running version post-restart:
7. SECRETS REVIEW
Unexplained crashes or restarts since exposure window? Y/N
Secrets resident in process memory:
Rotation decision: [rotate | not required, rationale]
Customization Guidance
Section 1, Priority Tier: Rank by the potential impact of a breach, not internet exposure. Identity systems and software-delivery infrastructure belong in Tier 1, as they can yield valuable tokens and credentials. Tier 2 includes collaboration and ticketing tools. Tier 3 covers everything else.
Section 2, Discovery Source: Go beyond your CMDB or SBOM. Runtimes often hide within commercial products, container images, and specific directories. Conduct a discovery pass to find Java binaries and reconcile them against your inventory. The gap reveals your findings.
Section 3, Hardware Context: In mixed environments, note if autoscaling can shift workloads between processor types. Treat "varies by node" as exposed until resolved. A hardware refresh can introduce exposure without application changes.
Section 4, Outbound Paths: Don't focus solely on internet exposure. Java programs can be compromised by connecting to malicious servers. Include outbound connections in your vulnerability-response procedure.
Section 6, Vendor Ownership: Each distribution needs its own fix. If the vendor controls the runtime, open a support case for the fixed version instead of patching around it.
Section 5, Severity: Don't rely solely on scores. Your context, including the secrets and access involved, should determine urgency.
Validation Steps
- Confirm the Running Version: After patching, ensure the process uses the fixed runtime. Record the post-restart version in Section 6.
- Verify CPU Capability Claims: Attach evidence like command output or node inventory references. Record unproven "no" as "unknown."
- Spot-Check Discovery: Randomly check commercial products and container images for embedded runtimes. Expand discovery if you find any missed by your inventory.
- Test Outbound Assumptions: For Tier 1 or 2 workloads, verify actual outbound destinations from network logs.
- Close the Secrets Loop: If exposed systems had unexplained crashes, review in-memory secrets and decide on rotation.
- Report Unknowns: Present leadership with workload status counts, including Unknown. A worksheet showing gaps is more honest than one that doesn't.
Start with identity and delivery systems, then expand. The goal isn't a perfect inventory, but knowing which answers you can prove.





