PHP Warns of Critical Security Flaws Affecting All Supported Versions, Fixes Coming November 19 · 5h 10-Year-Old NGINX HTTP/2 Flaw Lets Unauthenticated Attackers Execute Code on Servers · 15h F5 Warns 6 Critical BIG-IP Flaws Can Crash Traffic in Specific Setups, May Allow RCE · 15h CVE-2026-123633 and CVE-2026-123007: Critical NGINX HTTP/2 Vulnerabilities · 16h Critical Apache Roller RCE Flaw Under Active Exploitation, No Authentication Required · 17h Java4Shell Flaw Threatens Global Enterprises, Java Application Ecosystem · 17h How Cloudflare WAF helps protect against Java4Shell (CVE-2026-114938) · 19h Vulnerabilities in LMS software · 21h New Attack Bypasses TPM Lockout to Crack Windows Hello and BitLocker PINs in Days, Not Years · 21h AI Pentesting Checklist for Web Apps: 15 Questions to Ask Before Choosing a Vendor · 21h PHP Warns of Critical Security Flaws Affecting All Supported Versions, Fixes Coming November 19 · 5h 10-Year-Old NGINX HTTP/2 Flaw Lets Unauthenticated Attackers Execute Code on Servers · 15h F5 Warns 6 Critical BIG-IP Flaws Can Crash Traffic in Specific Setups, May Allow RCE · 15h CVE-2026-123633 and CVE-2026-123007: Critical NGINX HTTP/2 Vulnerabilities · 16h Critical Apache Roller RCE Flaw Under Active Exploitation, No Authentication Required · 17h Java4Shell Flaw Threatens Global Enterprises, Java Application Ecosystem · 17h How Cloudflare WAF helps protect against Java4Shell (CVE-2026-114938) · 19h Vulnerabilities in LMS software · 21h New Attack Bypasses TPM Lockout to Crack Windows Hello and BitLocker PINs in Days, Not Years · 21h AI Pentesting Checklist for Web Apps: 15 Questions to Ask Before Choosing a Vendor · 21h
🛡️
ACyber Sentinel Security Assessment Team
← Back to feed
⚡ Intelligence Report

PHP2SHELL Detection Coverage in THOR

PHP2SHELL Detection Coverage in THOR

THOR applies Sigma rules to supported log data on an endpoint during a scan. For PHP2SHELL, this means looking for exploit-related artifacts in web-server logs and for recorded post-exploitation activity. Private YARA rules provide additional checks for artifacts in log content.

The vulnerabilities, CVE-2026-109618 and CVE-2026-109619, are documented in PHP’s advisories on out-of-memory corruption and stale global state. Our PHP2SHELL work adds specific log indicators and process patterns to behavioral coverage that was already in place.

An exploit attempt is not proof of compromise

We distinguish between failed attempts and successful exploitation wherever the available evidence supports it. Some findings, however, remain signals that require investigation.

A log entry may suggest that an exploit attempt reached a vulnerable code path without showing whether the attacker achieved code execution. An isolated memory-exhaustion error or worker crash is weaker evidence: it can also have benign causes and does not independently establish the service’s patch status.

A clearly malicious reverse-shell command launched by PHP is much stronger evidence of code execution than an error message. Suspicious child processes still need context, and linking them to PHP2SHELL requires correlation with the surrounding activity. Other artifacts can establish compromise too.

Existing coverage for post-exploitation activity

After gaining code execution, attackers often use familiar commands and tools. They run reconnaissance, launch reverse shells, download payloads or establish persistence. The vulnerability used for entry may be new; these actions frequently are not.

Huntress’s PHP2SHELL write-up describes the same expected behavior: PHP workers spawning shells, interpreters, download tools and reconnaissance commands. It draws on previous PHP intrusions, including an example explicitly identified as unrelated to PHP2SHELL.

Nextron has developed generic detections for this activity over many years. The following Sigma rules were already in place before the PHP2SHELL vulnerabilities were published. They provided coverage for relevant post-exploitation behavior when the required events and rule conditions were present.

Pre-existing Sigma rule Availability and platform Coverage already present before PHP2SHELL
Linux Webshell Indicators Public, Linux; extended in PR #6506 Web-server processes directly spawning reconnaissance and system utilities such as whoami, hostname, ifconfig and netstat.
Linux Webserver Process Spawning Potential Reverse Shell Private, Linux Potential reverse-shell activity associated with web-server processes.
Web Server Executing Suspicious Commands Private, Linux Suspicious command execution associated with web-server processes.
Suspicious Process By Web Server Process Public, Windows; extended in PR #6506 Web-server processes spawning suspicious children, including cmd.exe, powershell.exe, net.exe, certutil.exe, rundll32.exe and wmic.exe.
Webshell Hacking Activity Patterns Public, Windows; extended in PR #6506 Webshell-related credential dumping, Active Directory reconnaissance, lateral movement and PowerShell download patterns.
Webshell Detection With Command Line Keywords Public, Windows; extended in PR #6506 Reconnaissance commands and utilities, suspicious shell command lines, and encoded or hidden PowerShell activity under matching web-server parents.
Webshell Tool Reconnaissance Activity Public, Windows; extended in PR #6506 Web-server processes checking for available tools with commands such as perl --help, python -h and wget --help.
Apache Segmentation Fault Public, Apache error logs; renamed and updated in PR #6506 Generic Apache worker segmentation-fault messages. This rule has existed since 2017 and can also match crashes unrelated to exploitation.

The Windows rules describe our broader coverage for post-exploitation activity across supported environments. The Linux Webshell Indicators extension and the new PHP2SHELL log rules are described below. Generic detections for reverse shells, reconnaissance and suspicious commands were already available before this work began.

THOR can apply these Sigma detections during a scan when the relevant process activity is present in supported endpoint logs. The process events need to contain the fields used by the rule, such as parent processes and command lines. Ordinary web access logs do not provide that process history.

In recent incidents, including NetScaler compromises, existing generic rules allowed us to detect post-exploitation activity on day one using detections that were already in place.

Based on past outbreaks, we estimate that existing generic rules can cover roughly 50 to 70 percent of the early post-exploitation activity we encounter, depending on the campaign and available evidence. This is an experience-based estimate, not a measured PHP2SHELL detection rate.

PHP2SHELL-specific additions

Our public contribution in SigmaHQ PR #6506 proposes updates to six existing rules and adds one new rule. The PR is open as of 22 October 2026.

For Linux Webshell Indicators, the update adds inline shell patterns observed in the PHP2SHELL PoC, including apache2 spawning sh -c "id" or sh -c "whoami". It also expands direct child-process coverage to download tools and interpreters and broadens the parent-process list, including explicit /php-fpm and /php-cgi matches. Coverage depends on the actual executable paths, process chain and command line recorded in the logs.

The four Windows process rules retain their existing suspicious-activity selectors and expand their parent-process lists with additional Apache, PHP and Tomcat executable names.

Log coverage combines one new rule with the existing Apache crash detection:

Sigma rule in PR #6506 Status Coverage
PHP Memory Corruption Log – Potential Php2shell Indicators New PHP fatal memory-exhaustion errors referencing rfc1867.c or an unknown source context, such as Unknown on line 0; session-open warnings containing /ndary=, sess_ and the other required warning terms.

These rules help identify activity worth investigating, including attempts whose outcome cannot be determined from the error logs alone. PoC-specific fragments may also change in other exploit implementations.

Our private YARA rules for PHP2SHELL-related log artifacts and suspicious PHP activity include:

  • LOG_PHP2SHELL_Artifacts_Oct26_1
  • LOG_PHP2SHELL_Artifacts_Oct26
  • LOG_SUSP_PHP2SHELL_Artifacts_Oct26
  • LOG_PHP_Exploitation_Indicators_Oct26

Logging determines what a scan can find

Apache’s standard common and combined access-log formats do not record the Content-Type or Cookie headers. Multipart boundaries or session identifiers carried only in those headers will not be available there. Request-level matching needs a source that actually captures the relevant values. See the Apache logging documentation.

Error logs are different. PHP may emit an upload-processing or session-path error containing a useful artifact without the server recording complete request headers. Check the effective log_errors and error_log settings, Apache’s LogLevel, and whether the resulting logs are available to THOR. The PHP logging documentation explains how PHP errors are routed.

THOR can examine retained evidence during a later scan. It cannot reconstruct process events or request details that were never logged or have already been discarded.

PHP hardening and temporary mitigations

As of 22 October 2026, we have received reports from multiple trusted sources that exploitation remains possible after the initial fixes. Continue installing PHP and distributor security updates, and review the following configuration mitigations for your deployment. Their effectiveness depends on the exploitation path. None provides complete protection against every variant.

The recommendations below draw on PHP’s advisories for out-of-memory corruption and stale global state:

  • Set memory_limit=-1 to remove the controllable PHP memory-limit failure used by confirmed exploitation paths. Enforce it administratively, for example with php_admin_value in the FPM pool or Apache mod_php configuration, and verify that no pool, directory or application setting restores a finite limit. An underlying allocation failure can still reach the vulnerable path. This also removes PHP’s per-request memory safeguard, increasing exposure to memory exhaustion and denial of service. Pair it with request limits. Follow the advisory when choosing resource-isolation controls: it specifically warns against RLIMIT_AS and RLIMIT_DATA, which can force the allocation failure that reaches the vulnerable path.
  • Keep post_max_size low. As application-dependent starting points, use 8M where file uploads are unnecessary and 32M where uploads are required, with upload_max_filesize below post_max_size. These values are not guaranteed safe thresholds. Enforce decoded request-size limits at the web server or proxy, including requests without a usable Content-Length. Also constrain input-field count, multipart-part count, headers per part and nesting depth at a layer that can enforce them. Where multipart form data is unnecessary, block it before it reaches PHP.
  • Set session.upload_progress.enabled=0 in php.ini, an FPM pool or server configuration that takes effect before request parsing. Do not rely on application ini_set() calls or FPM .user.ini files for this mitigation. Restart every worker. This blocks the confirmed session upload-progress paths, while other lifecycle variants remain possible.
  • On PHP 8.5, set fatal_error_backtraces=0 administratively and restart every worker. This blocks the confirmed fatal-backtrace path only.
  • Remove invalid or deprecated settings applied per request that emit diagnostics before initialization. Suppressing the resulting messages with error_reporting, display_errors, log_errors or @ does not prevent the vulnerable state access.

After applying changes, restart every PHP worker and verify the effective configuration for each serving pool and virtual host. Consult PHP’s FPM configuration reference and core configuration reference when implementing the settings.

This section records our guidance as of 22 October 2026. We will not maintain this article as a continuously updated mitigation guide. Follow the latest PHP security advisories, your distributor’s updates and current researcher write-ups. Researchers also publish new findings and recommendations on X/Twitter; follow their linked technical analyses as the understanding of these vulnerabilities develops.

What customers should do

The next regular signature release is scheduled for the morning of 23 October 2026 and will include PHP2SHELL detections. Customers can already use the new detections through the preview signature channel.

For standalone THOR, run thor-util update --sigdev on Linux or macOS, or thor-util.exe update --sigdev on Windows. See the THOR Util documentation.

In Management Center v4, enable Show Signature Preview Channel under Advanced Settings, save the change, and select the Preview channel for THOR signatures through Version Pinning. In ASGARD Management Center v3, the setting is called Show Signature SigDev Option; enable it and select the preview signatures when creating a scan. See the instructions for v4 or v3.

Preview signatures have passed automated checks but are still undergoing manual validation, so additional false positives are possible.

Preserve relevant logs and ensure that supported PHP, web-server and process-event logs are included in the scan. Include retained logs from before patching where available. Remediation does not remove the need to investigate earlier activity.

Treat exploit-related matches according to what they actually show. Correlate them with malicious process execution, outbound connections, payload files and persistence artifacts before drawing conclusions about the outcome.

An absence of findings does not rule out compromise. Exploitation can remain inside a worker process, and PHP’s advisory warns that affected paths may lack predictable log entries. Follow the current PHP remediation guidance regardless of the scan result.

About the author:

Avatar photo

Nextron Threat Research Team

The Nextron Threat Research Team builds the detection logic behind THOR, Aurora, Thunderstorm and the rest of the Nextron toolchain. The group analyses intrusions, reverse-engineers malware, tracks supply-chain incidents, and turns all of that into signatures, heuristics and rules used across our products. The team maintains YARA and Sigma content at scale, develops internal tooling and pipelines, and ships thousands of high-quality detections every year that help customers spot real attacker activity instead of noise.

Referenced Vulnerabilities

CVE-2026-109619 CVSS 9.8 CRITICAL Modified
CVE-2026-109619 - RCE via stale global state
PHP has various global state that temporarily points into request memory. In some scenarios, global state may not be correctly reset, or may be repopulated after being reset. The global state may then refer to request memory even after it has been freed. Subsequent requests may read or branch on this stale state before it is fully initialized. Multiple strategies can trigger this condition and produce remote code execution (RCE).
CVE-2026-109618 CVSS 9.8 CRITICAL Analyzed
CVE-2026-109618 - RCE via OOM corruption
PHP uses longjmp() for out-of-memory (OOM) errors to exit the current context and abort handling of the request. Some contexts may not properly account for this scenario and leave request memory in an uninitialized state before performing an allocation. The subsequent request cleanup procedure may read or branch on this memory. Multiple OOM strategies can consistently trigger this condition and lead to remote code execution (RCE).
CVE-2026-88772 CVSS 9.5 CRITICAL Analyzed
CVE-2026-88772 - Citrix NetScaler Improper Restriction of Operations within the Bounds of a Memory Buffer Vulnerability - [Actively Exploited]
Vulnerability in Citrix NetScaler ADC and Citrix NetScaler Gateway. This issue affects ADC: before 14.1-73.37, before 13.1-64.23, before 14.1-73.37 FIPS, and before 13.1.37.279 FIPS and NDcPP; Gateway: before 14.1-73.37 and before 13.1-64.23 leading to Remote Code Execution or Denial of Service
CVE-2026-88771 CVSS 9.8 CRITICAL Analyzed
CVE-2026-88771 - Citrix NetScaler Improper Input Validation Vulnerability - [Actively Exploited]
Improper input validation vulnerability in Citrix NetScaler ADC and Citrix NetScaler Gateway. This issue affects ADC: before 14.1-73.37, before 13.1-64.23, before 14.1-73.37 FIPS, and before 13.1.37.279 FIPS and NDcPP; Gateway: before 14.1-73.37 and before 13.1-64.23 leading to an unauthenticated attacker to execute arbitrary commands.
CVE-2026-54121 CVSS 8.8 HIGH Analyzed
CVE-2026-54121 - Active Directory Certificate Services Elevation of Privilege Vulnerability
Improper authorization in Active Directory Certificate Services (AD CS) allows an authorized attacker to elevate privileges over a network.