Java4Shell & NGINX2Root: Find exposures, monitor post-exploit activity, and remediate faster with CrowdStrike guidance. Learn more
Critical RCE CVEs Resource Center

Publicly Disclosed Critical RCE CVEs: CrowdStrike Research and Guidance

Uncover exposure, monitor for post-exploit activity, and accelerate remediation of Java4Shell and NGINX2Root with CrowdStrike research and guidance.

Get expert analysis, technical findings, and guidance for your next steps.

Blog
Java4Shell (CVE-2026-114938): CrowdStrike Detects Exposure and Post-Exploitation Activity

Learn how to understand your exposure and detect post-exploitation activity related to Java4Shell.
Demo
Java4Shell: Exposure and Response with the CrowdStrike Falcon Platform

See how the CrowdStrike Falcon® platform helps identify potential exposure, prioritize remediation, and patch affected systems across endpoints and cloud environments.
Free Trial
Try the Falcon Platform Free

Start your free trial to see how Falcon finds exposures and automates protection.

Frequently Asked Questions

Java4Shell (CVE-2026-114938) is a memory corruption vulnerability in the HotSpot Java Virtual Machine (JVM), the runtime shared by Oracle Java SE, OpenJDK, and many downstream Java distributions. The flaw is in HotSpot's AVX-512-optimized Base64 decoding routine, which the JVM uses only on processors that support AVX-512 VBMI. An unauthenticated remote attacker who can pass a specially crafted Base64 string to an affected application can trigger an out-of-bounds heap write of up to 62 bytes.

Only x86-64 processors that support AVX-512 VBMI are affected (Intel Ice Lake and newer, AMD Zen 4 and newer). ARM64 and i386 systems are not affected.

 

Branch

Affected versions

Fixed version

Java 17

17.0.6 through 17.0.21

17.0.21.1

Java 21

Up to 21.0.13

21.0.13.1

Java 25

Up to 25.0.5

25.0.5.1

Java 27

Up to 27.0.1

27.0.1.1

Oracle Java 8 Performance Pack

8u511b33-perf

See Oracle advisory

 

Standard Java 8 and Java 11 environments are not affected; the Oracle Java 8 Performance Pack is the exception. Downstream distributions are also affected, including Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu, Red Hat OpenJDK, Oracle GraalVM, BellSoft Liberica, and SAP SapMachine. Applications that bundle their own Java runtime, container images, and managed cloud services can be affected as well.

Exposure depends on whether an attacker can deliver a crafted Base64 string to an application running an affected runtime on supported hardware. Common entry points include authentication headers, tokens (such as JWTs and SAML assertions), and file uploads, which are often processed before authentication completes. Internet-facing services are the highest priority and should be patched first. Internal systems that meet the same conditions remain vulnerable and can be addressed through regular scheduled patching.

The most likely outcome of exploitation is denial of service, so unexplained JVM crashes on affected hosts are worth investigating. Memory disclosure ("JavaBleed") and remote code execution are possible but harder to achieve. Teams should look for post-exploitation activity such as web shell deployment, suspicious child processes spawned by Java processes, hands-on-keyboard activity, and living-off-the-land techniques.

Falcon Exposure Management uses sensor-based assessment to identify Java runtimes affected by CVE-2026-114938 on supported Windows, Linux, and macOS hosts. Coverage includes Oracle Java SE, OpenJDK, and downstream distributions such as Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu, Red Hat OpenJDK, Oracle GraalVM, BellSoft Liberica, and SAP SapMachine. Hosts updated to 17.0.21.1, 21.0.13.1, 25.0.5.1, or 27.0.1.1 are no longer flagged.

Standard install checks work on all supported sensor versions. Additional checks find custom or bundled installs: a system-wide Java check on Windows (sensor 7.40 or later, with "Enable Full System Visibility" turned on, C: drive only) and a running-process check on Linux (sensor 7.31 or later). On Linux, Exposure Management also checks whether AVX-512 VBMI is enabled. On Windows and macOS it may flag hosts that are not exploitable, so prioritize systems with supported processors.

Falcon Exposure Management and Falcon Spotlight customers can find exposed hosts in the Falcon console under Exposure Management > Vulnerability Management > Vulnerabilities. For containers, Falcon Cloud Security image assessment identifies images carrying an affected runtime and shows which are running. Falcon Cloud Security Cloud Risk flags internet-exposed instances running an affected Java version.

Several conditions must all be true at once: an affected Java version, an x86-64 processor with AVX-512 VBMI, and an application that decodes attacker-supplied Base64. The overwrite is limited to 62 bytes, and what it corrupts depends on what sits next to the buffer on the heap. Heap layout varies across JVM versions, garbage collectors, and workloads. Denial of service is the most reliable outcome, and reliable remote code execution is the hardest to achieve, though not impossible.

If an attacker achieved code execution, activity could include web shell deployment, suspicious child processes spawned by Java processes, hands-on-keyboard activity, and living-off-the-land techniques. Memory disclosure could expose fragments of other requests, session tokens, or credentials.

The Falcon platform provides out-of-the-box visibility and detection for post-exploitation activity that may follow exploitation of CVE-2026-114938. Existing indicators of attack (IOAs) detect and prevent common post-exploitation behavior, including web shell deployment, suspicious child processes spawned by Java processes, hands-on-keyboard activity, and living-off-the-land techniques. CrowdStrike researchers are analyzing potential exploitation vectors to develop new behavioral detection logic.

Patch systems that meet all three conditions immediately, starting with internet-facing Java services. Apply the November 2026 fixes from Oracle and OpenJDK or the equivalent update from your Java vendor. Then:

  • Restart all running JVMs. Upgrading the runtime on disk does not protect processes that are already running.

  • Rebuild container images on patched base images and redeploy them.

  • Inventory embedded runtimes in appliances and third-party applications, and check vendor advisories and the CISA alert on CVE-2026-114938.

Falcon for IT Application Management can deploy Java updates at scale for runtimes available in its Application Catalog. If a runtime is missing, update it with the vendor's installer through a Falcon for IT custom task or existing deployment tooling. It can take up to 24 hours for hosts to report on newly managed applications.

Blog
NGINX2Root: Identify and Protect with the CrowdStrike Falcon Platform

Discover how the CrowdStrike Falcon® platform helps identify exposed assets and detect post-exploitation activity.
Free Trial
Try the Falcon Platform Free

Start your free trial to see how Falcon finds exposures and automates protection.

Frequently Asked Questions

NGINX2Root refers to five critical vulnerabilities in nginx that can allow unauthenticated remote attackers to execute malicious code. When chained with a separate Linux kernel vulnerability, attackers may escalate privileges to root and escape containers to compromise the underlying host.

NGINX2Root was publicly disclosed on November 18, 2026. The five vulnerabilities are tracked as CVE-2026-123007, CVE-2026-123634, CVE-2026-123633, CVE-2026-121810, and CVE-2026-113425. The two most severe vulnerabilities carry CVSS 4.0 scores of 9.2 (Critical).

The vulnerabilities affect different nginx components, including HTTP/2 and HTTP/3 request handling, reverse proxy configurations, and mail proxy functionality. Specially crafted network traffic can enable remote code execution. A separate Linux kernel flaw may then allow attackers to escalate privileges to root.

Multiple versions of nginx Open Source and nginx Plus are affected, depending on the vulnerability and configuration. Fixed releases include nginx Open Source 1.31.9 (mainline) and 1.30.7 (stable), and nginx Plus 37.3.0.1 and 37.0.8.1. Organizations should consult their vendor's security advisory for applicable updates.

No. Exploitability depends on the nginx version and configuration. For example, CVE-2026-123007 and CVE-2026-123633 require HTTP/2 to be enabled, while other vulnerabilities depend on specific proxy, resolver, HTTP/3, or mail proxy configurations.

Successful exploitation can enable remote code execution and unauthorized access to affected systems. When combined with a Linux kernel privilege-escalation vulnerability, attackers may gain root access and escape Docker containers to compromise the underlying host.

Organizations should identify affected nginx installations, prioritize internet-facing systems, and apply vendor-provided security updates. A full nginx restart is required after patching. Disabling HTTP/2 can temporarily mitigate CVE-2026-123007 and CVE-2026-123633, but does not address the other three vulnerabilities.

Organizations can use external attack surface management (EASM) to discover nginx services exposed to the internet. CrowdStrike Falcon Exposure Management provides visibility into known internet-facing nginx assets, helping security teams identify potential exposure and prioritize further investigation.

How does CrowdStrike help organizations address NGINX2Root?

 

The CrowdStrike Falcon platform helps organizations identify vulnerable systems, prioritize remediation, and detect suspicious post-exploitation activity.

  • Falcon Exposure Management: Identifies affected nginx installations through host-based vulnerability assessment, network scanning, and EASM.

  • Falcon Cloud Security: Identifies vulnerable container images and affected running containers.

  • Falcon for IT: Provides inventory, exposure assessment, and Linux remediation workflows through a dedicated NGINX2Root content pack.

  • Falcon Endpoint Security: Detects suspicious post-exploitation behaviors, including malicious processes, web shells, and credential access.

CrowdStrike provides behavioral detection and prevention capabilities for suspicious post-exploitation activity, including malicious child processes, web shell activity, and unauthorized access to sensitive files. Because the initial exploit occurs within nginx worker memory, endpoint detection focuses on post-exploitation behavior. These capabilities complement, but do not replace, patching affected nginx installations.
Webinar
PHP2Shell webinar with CrowdStrike and OpenAI

Learn how to identify affected systems, investigate suspicious activity, and reduce PHP2Shell risk.
Blog
CrowdStrike Protects Customers from Threats Delivered via PHP2Shell

Explore CrowdStrike's technical findings, exploitation paths, and detection and protection guidance.
Demo
Falcon Protection Against PHP2Shell

See how the CrowdStrike Falcon® platform helps identify affected systems, find exposures, and investigate potential exploitation.
In the News
George Kurtz on CNBC: PHP Vulnerability

Watch CrowdStrike CEO & Founder George Kurtz on CNBC discuss PHP2Shell and potential risks.
Blog
Obstacles to Weaponizing PHP2Shell Likely Constrain Immediate Widespread Exploitation

Learn why obstacles to weaponizing PHP2Shell may limit immediate widespread exploitation.
Free Trial
Try the Falcon Platform Free

Start your free trial to see how Falcon finds exposures and automates protection.

Frequently Asked Questions

PHP2Shell comprises two new vulnerabilities, CVE-2026-109618 and CVE-2026-109619.


These vulnerabilities are not a bypass of CVE-2024-4577. They do not share the same root cause or execution code path, meaning that patching CVE-2024-4577 does not address PHP2Shell.

Yes. Internet-facing systems generally have greater direct exposure, but internal PHP services can still be reached by an attacker who has gained access to the network. Assess internal and external assets based on reachability, affected versions, configuration, and business criticality.

CrowdStrike experts recommend reviewing the published PHP2Shell technical research for illustrated process-tree behavior and examples of detection and prevention activity.
 

Organizations should use the published technical research and available security telemetry to guide their investigations.

Falcon Exposure Management supports discovery of internet-facing PHP services through External Attack Surface Management (EASM) and identification of PHP installations on supported managed systems through asset and vulnerability assessment capabilities. Coverage varies by discovery method and deployed sensor version. These findings help teams determine which assets require further investigation and remediation.

As of October 28, 2026, CrowdStrike had observed probing and attempted exploitation but had not confirmed successful exploitation in the wild. This describes the assessment at that time; exploitation status can change as new evidence emerges.

A proof-of-concept exploit may depend on PHP configuration, worker behavior, memory conditions, and server architecture. Those conditions vary across deployments, making reliable exploitation of unknown targets more difficult than reproducing an exploit in a controlled environment. Changes in exploit reliability or attacker tooling could alter the threat.

Potential signs of post-exploitation activity include unexpected child processes launched by PHP or web-server processes, suspicious command execution, web-shell activity, or unusual outbound connections. These indicators require investigation in context; their absence does not establish that exploitation has not occurred. CrowdStrike's PHP2Shell technical research includes behavioral examples.

Falcon Next-Gen SIEM provides PHP-related hunting content in the Emerging Threat Response dashboard, accessible through Log Management > Dashboards. Available investigations depend on the telemetry ingested and the detection content deployed.
 

A Falcon Next-Gen SIEM correlation template is available under the name “Generic - Web - PHP Memory Exhaustion During Request Startup (CVE-2026-109618 & CVE-2026-109619).” Its effectiveness depends on relevant log sources, field mappings, and configuration. Support for a particular web-server deployment, including IIS with FastCGI, should be confirmed against the current content documentation.

Falcon can detect and, with applicable recommended prevention policies, block certain malicious post-exploitation behaviors. Protection depends on the observed behavior, deployed capabilities, and policy configuration; it does not guarantee prevention of every exploitation path or malicious action.

Use risk-based compensating controls while planning the vendor-supported update. Depending on the deployment, measures may include restricting access to the affected service, reducing unnecessary exposure, and monitoring for suspicious activity. Individual mitigations may block only specific exploitation paths and do not replace the security fix.

Charlotte AI can assist analysts with security investigations, including exploring potentially affected assets, developing hunting queries, and reviewing vulnerability context available through connected Falcon capabilities. Results require validation against underlying telemetry and the customer's enabled features.

AI-assisted security analysis can be incorporated into development and CI/CD workflows to identify potential weaknesses earlier. Its findings should be validated through established engineering, testing, and security review processes.