---
title: >-
  PHP2Shell session memory exhaustion with session decode failure or upstream
  crash on nginx
description: Datadog, the leading service for cloud-scale monitoring.
breadcrumbs: >-
  Docs > Datadog Security > OOTB Rules > PHP2Shell session memory exhaustion
  with session decode failure or upstream crash on nginx
---

> For the complete documentation index, see [llms.txt](https://docs.datadoghq.com/llms.txt).

# PHP2Shell session memory exhaustion with session decode failure or upstream crash on nginx

{% alert level="danger" %}
This rule is part of a beta feature. To learn more, [contact Support](https://docs.datadoghq.com/help/).
{% /alert %}
Classification:attackTactic:[TA0001-initial-access](https://attack.mitre.org/tactics/TA0001)Technique:[T1190-exploit-public-facing-application](https://attack.mitre.org/techniques/T1190) 
## Goal{% #goal %}

Detects PHP2Shell (CVE-2026-109618/109619) exploitation stages on an nginx host running PHP-FPM. A pre-script PHP memory exhaustion accompanied by a session subsystem error, a session decode failure, or an upstream worker crash on the same host and service signals an exploit attempt.

## Strategy{% #strategy %}

This rule monitors nginx error logs for PHP messages relayed from PHP-FPM workers. PHP2Shell is a memory-safety vulnerability in PHP's multipart and session upload-progress handling that leads to remote code execution.

- In the **arm** stage, a large multipart `POST` exhausts `memory_limit` before any script runs, so PHP logs `Allowed memory size of ... exhausted` with the location `in Unknown on line 0` instead of a script path. A `Failed to decode session object` warning often follows.
- In the **fire** stage, a second request crashes the PHP-FPM worker, and nginx logs `upstream prematurely closed connection`. Ordinary out-of-memory errors name the script and line that ran out of memory, so a pre-script memory failure is rare on its own.

The rule raises signals with different levels of severity, depending on the situation:

- `Critical`: The memory failure occurs on the same host and service, within five minutes, as a PHP session subsystem error that was raised before any script ran. This error can be a failure to read or write session data, or a rejected session ID, each ending `in Unknown on line 0`. This pairing matches crash-free exploit chains that keep the worker alive and never corrupt cross-request session state, so they produce neither a worker crash nor a session decode failure.
- `High`: The memory failure and the session decode failure occur together.
- `Medium`: The memory failure and an upstream worker crash occur together.

## Triage & Response{% #triage--response %}

- Examine the raw nginx error lines for the memory failure and the crash on `{{host}}`, and extract the `client:` address and `request:` line embedded in the `FastCGI sent in stderr` message.
- Review nginx access logs on `{{host}}` around the signal for large multipart `POST` requests from that client, especially requests that returned `500` or `502`.
- Determine whether the memory failure came from a legitimate oversized upload by checking the request path, the client, and whether `{{service}}` normally accepts large file uploads.
- Review any PHP session subsystem errors (`Failed to read session data`, `Failed to write session data`, or `Session ID is too long or contains illegal characters`) on `{{host}}`, and distinguish a misconfigured `session.save_path`, which raises these warnings on their own, from warnings that appear only in the same request as the memory failure.
- Check whether the PHP-FPM workers on `{{host}}` were restarted or crashed repeatedly, and compare the timing with the requests from the same client.
- Identify the PHP version running on `{{host}}` and verify whether it includes the fix (8.5.12, 8.4.27, 8.3.36 or 8.2.35).
- Inspect the host for signs of successful exploitation, such as child processes spawned by PHP-FPM workers, in-memory executables, new files owned by the web server user, or unexpected outbound connections.
