Back to Intelligence

CVE-2025-42884: SAP NetWeaver Enterprise Portal JNDI Injection Added to CISA KEV — Detection and Remediation Guide

SA
Security Arsenal Team
November 20, 2026
10 min read

On November 20, 2026, CISA added CVE-2025-42884 to the Known Exploited Vulnerabilities (KEV) catalog, formally confirming what many of us in incident response have suspected for weeks: SAP NetWeaver Enterprise Portal is being actively exploited in the wild via an improper neutralization of special elements in data query logic vulnerability — in practical terms, a JNDI injection flaw.

The vulnerability allows an unauthenticated, remote attacker to disclose or modify server information by injecting malicious JNDI environment properties or a crafted URL that the server then uses during JNDI lookup operations. If you've been in this industry since Log4Shell, the attack pattern will feel uncomfortably familiar: coerce the server into resolving an attacker-controlled directory lookup, and you own a path into the JVM.

SAP systems sit at the heart of financial, HR, logistics, and manufacturing operations for the majority of the Fortune 500. They hold crown-jewel data, they are frequently under-patched due to change-control friction, and they are persistently targeted by both financially motivated actors and nation-state operators. A KEV listing for an unauthenticated NetWeaver flaw is a drop-everything event for any organization running the affected component.

Bottom line up front: Treat this as an emergency change. Apply SAP's mitigations per the vendor security note, comply with CISA's BOD 26-04 prioritization and Forensics Triage requirements, and hunt your perimeter and SAP Java stacks for signs of prior compromise before you assume the patch closed the door.


Technical Analysis

Affected Component

  • Product: SAP NetWeaver Enterprise Portal (Java stack)
  • Vulnerability class: Improper neutralization of special elements in data query logic (CWE-943 family), manifesting as JNDI injection
  • Attack vector: Network, unauthenticated, low complexity
  • Impact: Unauthorized disclosure and modification of server-side information; depending on the JDK version and available gadget classes in the classpath, JNDI injection against Java applications has historically been escalable to remote code execution

How the Vulnerability Works

The Java Naming and Directory Interface (JNDI) is an abstraction layer that lets Java applications look up resources — datasources, message queues, remote objects — from directory services such as LDAP, RMI, DNS, and CORBA. The flaw in Enterprise Portal is that attacker-controllable input reaches JNDI lookup logic without adequate sanitization of special elements in the query.

From a defender's perspective, the exploitation chain looks like this:

  1. Reconnaissance: The attacker identifies an exposed SAP Enterprise Portal instance (typically HTTP/HTTPS on ports 50000, 50001, 443, or 8443). Shodan and Censys make this trivial.
  2. Injection: The attacker submits a crafted request containing JNDI environment properties or a lookup URL (e.g., ldap://attacker-controlled-host:1389/... or rmi://...) into the vulnerable data query parameter.
  3. Lookup coercion: The vulnerable Portal component passes the attacker-supplied value into InitialContext.lookup() or equivalent, causing the SAP JVM to initiate an outbound connection to an attacker-controlled directory service.
  4. Post-exploitation: At minimum, the attacker extracts or modifies server information. Where the runtime permits remote codebase loading or exploitable deserialization gadgets exist, this becomes a foothold for code execution and webshell deployment on the NetWeaver host.

Exploitation Status

  • CISA KEV: Confirmed active exploitation as of 2026-11-20
  • Authentication required: None
  • Exposure: Any internet-facing or internally reachable Enterprise Portal instance is at risk; internal segmentation does not substitute for patching given lateral movement tradecraft

This is not theoretical. KEV inclusion means CISA has reliable evidence of in-the-wild exploitation. Assume scanning is widespread and indiscriminate.


Detection & Response

The most reliable high-fidelity signals for JNDI exploitation against a Java application server are: (1) the SAP Java process initiating outbound LDAP/RMI connections, and (2) JNDI protocol strings appearing in web request logs destined for the Portal. These are low-noise in almost every environment — legitimate SAP JVMs have no business making outbound LDAP calls to the internet.

Sigma Rules

YAML
---
title: SAP Java Process Outbound LDAP or RMI Connection
description: Detects the SAP NetWeaver Java stack (jlaunch / sapcontrol child JVM) initiating outbound connections on common JNDI service ports, consistent with CVE-2025-42884 JNDI lookup coercion.
references:
  - https://www.cisa.gov/known-exploited-vulnerabilities-catalog?search_api_fulltext=CVE-2025-42884
  - https://attack.mitre.org/techniques/T1190/
author: Security Arsenal
date: 2026/11/21
status: experimental
id: 3f9c1b72-6a4e-4d2b-9c51-7e2a8f40b311
tags:
  - attack.initial_access
  - attack.t1190
logsource:
  category: network_connection
  product: windows
detection:
  selection_process:
    Image|contains:
      - '\jlaunch.exe'
      - '\sapjvm'
      - '\java.exe'
  selection_ports:
    DestinationPort:
      - 389
      - 636
      - 1389
      - 1099
      - 1098
  filter_internal:
    DestinationIp|cidr:
      - '10.0.0.0/8'
      - '172.16.0.0/12'
      - '192.168.0.0/16'
  condition: selection_process and selection_ports and not filter_internal
falsepositives:
  - Rare legitimate directory integrations; verify against your documented SAP landscape before suppressing
level: high
---
title: JNDI Lookup String in Web Request Targeting SAP Portal
description: Detects inbound web requests containing JNDI protocol handler strings (ldap, rmi, ldaps, dns, iiop) characteristic of CVE-2025-42884 exploitation attempts against SAP NetWeaver Enterprise Portal.
references:
  - https://www.cisa.gov/known-exploited-vulnerabilities-catalog?search_api_fulltext=CVE-2025-42884
  - https://attack.mitre.org/techniques/T1190/
author: Security Arsenal
date: 2026/11/21
status: experimental
id: 8b2e4d19-1c7f-4a35-b6d2-9f0e3a5c7742
tags:
  - attack.initial_access
  - attack.t1190
logsource:
  category: webserver
detection:
  selection_uri:
    cs-uri-query|contains:
      - 'ldap://'
      - 'ldaps://'
      - 'rmi://'
      - 'iiop://'
      - 'dns://'
      - '%24%7Bjndi' # URL-encoded ${jndi
      - '${jndi:'
  condition: selection_uri
falsepositives:
  - Vulnerability scanners; tune by source IP against your approved scanning ranges rather than disabling
level: critical

Note on tuning rule two: deploy it against your reverse proxy / WAF / IIS logs fronting the Portal, or against SAP ICM access logs if you forward them to your SIEM. Source-IP filtering for known scanner ranges is the correct tuning approach — never blanket-suppress ${jndi patterns.

KQL (Microsoft Sentinel / Defender)

This hunt assumes you are ingesting SAP host Syslog, firewall/proxy logs via CommonSecurityLog, and Defender network events from any Windows-based SAP application servers. Hunt back at least 30 days — patching does not retroactively clean a compromise.

KQL — Microsoft Sentinel / Defender
let Lookback = 30d;
let JndiPorts = dynamic([389, 636, 1389, 1099, 1098]);
let JndiPatterns = dynamic(["ldap://", "ldaps://", "rmi://", "iiop://", "dns://", "${jndi", "%24%7Bjndi"]);
union isfuzzy=true
  (CommonSecurityLog
   | where TimeGenerated > ago(Lookback)
   | where DestinationPort in (JndiPorts)
   | where DeviceAction !in ("deny", "blocked")
   | where not(ipv4_is_private(DestinationIP))
   | project TimeGenerated, SourceIP, DestinationIP, DestinationPort, DeviceProduct, RequestURL=coalesce(RequestURL, ""), SourceProcessName
  ),
  (Syslog
   | where TimeGenerated > ago(Lookback)
   | where SyslogMessage has_any (JndiPatterns)
   | project TimeGenerated, Computer, ProcessName, SyslogMessage
  ),
  (DeviceNetworkEvents
   | where TimeGenerated > ago(Lookback)
   | where InitiatingProcessFileName has_any ("jlaunch", "java", "sapcontrol")
   | where RemotePort in (JndiPorts)
   | where not(ipv4_is_private(RemoteIP))
   | project TimeGenerated, DeviceName, InitiatingProcessFileName, InitiatingProcessCommandLine, RemoteIP, RemotePort
  )
| sort by TimeGenerated desc

Any hit on the third branch — an SAP JVM reaching out to a public IP on an LDAP/RMI port — should be treated as a probable successful exploitation and escalated to IR immediately. Correlate hits with the requesting source IP: that is your threat actor.

Velociraptor VQL

Use this artifact against Windows-based SAP application servers to surface JVM processes with suspicious command lines and active external connections. For Linux-based SAP hosts, pair this with the Bash triage script below.

VQL — Velociraptor
-- CVE-2025-42884 triage: SAP JVM processes and suspicious outbound connections
SELECT Pid, Name, CommandLine, Exe, Username, CreateTime
FROM pslist()
WHERE Name =~ '(?i)jlaunch|java|sapcontrol'

SELECT Pid, Name, RemoteAddress, RemotePort, Status
FROM netstat()
WHERE Name =~ '(?i)jlaunch|java'
  AND RemotePort IN (389, 636, 1389, 1099, 1098)
  AND RemoteAddress NOT LIKE '10.%'
  AND RemoteAddress NOT LIKE '192.168.%'
  AND RemoteAddress NOT LIKE '172.16.%'

Remediation and Verification Script

SAP NetWeaver Java stacks most commonly run on Linux. This script validates exposure by checking for listening Portal services, reviews ICM/HTTP access logs for JNDI indicators, and flags suspicious outbound connections from the SAP JVM. Run it on each SAP application server, then forward results to your IR channel.

Bash / Shell
#!/bin/bash
# CVE-2025-42884 SAP NetWeaver Enterprise Portal exposure & compromise triage
# Run as root or sidadm on each SAP Java application server

echo "=== [1] Listening SAP ICM/Portal services ==="
ss -tlnp | grep -E ':(50000|50001|443|8443|8000)' || echo "No standard SAP HTTP ports listening"

echo "=== [2] Active outbound connections from SAP JVM on JNDI ports ==="
ss -tnp | grep -E ':(389|636|1389|1099|1098)' | grep -E 'java|jlaunch' || echo "No suspicious JNDI-port egress found"

echo "=== [3] Scanning ICM/HTTP access logs for JNDI injection patterns (last 30 days) ==="
find /usr/sap -name '*.trc' -o -name 'access*' -o -name '*.log' 2>/dev/null | while read -r f; do
  find "$f" -mtime -30 -type f 2>/dev/null
done | xargs grep -lEi 'ldap://|ldaps://|rmi://|iiop://|dns://|\$\{jndi|%24%7Bjndi' 2>/dev/null | tee /tmp/cve-2025-42884_hits.txt || echo "No JNDI patterns found in logs"

echo "=== [4] SAP NetWeaver release info for patch verification ==="
cat /usr/sap/*/SYS/profile/DEFAULT.PFL 2>/dev/null | grep -i -E 'SAPSYSTEMNAME|SAPGLOBALHOST'
ls /usr/sap/*/*/j2ee/cluster/server*/apps/sap.com 2>/dev/null | head -20

echo "=== [5] Recently modified JSP/class files in webapps (webshell indicator, last 14 days) ==="
find /usr/sap -path '*irj*' \( -name '*.jsp' -o -name '*.class' \) -mtime -14 -type f 2>/dev/null | head -50

echo "Triage complete. Preserve /tmp/cve-2025-42884_hits.txt and full log copies per CISA Forensics Triage Requirements BEFORE patching."

For Windows-based SAP hosts, adapt the equivalent checks with Get-NetTCPConnection, Get-Process jlaunch,java, and Select-String against the ICM trace logs — the indicators are identical.


Remediation

  1. Apply SAP's mitigation immediately. Follow the SAP Security Note referenced in the CISA KEV entry and SAP's official advisory channel (support.sap.com → Security Notes). SAP delivers NetWeaver fixes via Support Packages and security notes; obtain the exact note number from the KEV catalog entry and your SAP Solution Manager, and apply per vendor instructions. Do not wait for your next maintenance window.
  2. Comply with CISA BOD 26-04. Federal civilian agencies are bound by the KEV due date; private-sector organizations should treat CISA's remediation deadline as their own. Where the product cannot be remediated, CISA's guidance directs you to discontinue use or apply compensating controls — take that language seriously.
  3. Honor CISA's Forensics Triage Requirements. Before patching, capture and preserve evidence: ICM/HTTP access logs, default traces, security audit logs, and memory or disk images of internet-facing nodes. Patching first destroys evidence of whether you were already breached — and with confirmed active exploitation, "were we hit?" is a question your board will ask.
  4. If you cannot patch today, reduce exposure:
    • Remove Enterprise Portal from direct internet exposure; place it behind a VPN or identity-aware proxy.
    • Deploy WAF/reverse-proxy rules blocking requests containing ldap://, ldaps://, rmi://, iiop://, dns://, and ${jndi (and URL-encoded variants) in URI, query string, and POST bodies.
    • Egress-filter the SAP application servers: the SAP JVM should be permitted outbound only to documented SAP infrastructure. Block 389/636/1389/1099/1098 egress to the internet at the perimeter.
    • Where the JDK version permits, verify JNDI remote codebase loading is disabled (com.sun.jndi.ldap.object.trustURLCodebase=false, com.sun.jndi.rmi.object.trustURLCodebase=false) — default in modern JDKs, but confirm on your SAPJVM.
  5. Assume compromise and hunt. Run the detections above across a minimum 30-day window. Check for newly created portal users, unexpected JSP files under IRJ webapps, and unusual child processes of the SAP JVM (a classic post-exploitation indicator is jlaunch/java spawning cmd.exe, powershell.exe, sh, or curl).
  6. Rotate credentials for any SAP service accounts, RFC destinations, and directory bindings stored on potentially exposed systems if any suspicious activity is found.

The Strategic Lesson

SAP's attack surface keeps producing KEV entries because the adversary math is simple: one successful SAP compromise can yield financial fraud, intellectual property theft, and a staging ground into everything the ERP touches. If your vulnerability management program treats SAP as a "quarterly patch cycle" asset, this CVE is your forcing function to reclassify internet-reachable SAP components as emergency-patch, Tier-0 assets — with the same SLAs you apply to your VPN concentrators and identity providers.

Related Resources

Security Arsenal Penetration Testing Services AlertMonitor Platform Book a SOC Assessment vulnerability-management Intel Hub

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.