---
id: "CVE-2026-106807"
kind: "vulnerability"
aliases: ["CVE-2026-106807"]
title: "Open5GS: PFCP transaction pool exhaustion crashes SMF/UPF, stopping session setup"
severity: "high"
cvss: null
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["5g-core", "open5gs", "runtime", "smf", "telecom", "upf"]
not_affected_if: ["not-open5gs-2.8.0", "pfcp-not-exposed"]
published_at: "2026-11-02T17:17:33+00:00"
first_seen_at: "2026-11-02T17:24:12+00:00"
updated_at: "2026-11-02T17:26:24+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-106807", "first_seen_at": "2026-11-02T17:24:12+00:00"}]
---

# Open5GS: PFCP transaction pool exhaustion crashes SMF/UPF, stopping session setup

Open5GS 2.8.0 mishandles a failed PFCP transaction allocation, which trips an assertion and aborts the network function process. An attacker who can send PFCP messages to the SMF or UPF can exhaust transaction resources and knock the service offline until it is restarted.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Run `open5gs-smfd --version` or check the installed Open5GS version on SMF and UPF hosts, expecting `2.8.0` to be affected.
- Confirm whether PFCP port 8805 is reachable from untrusted networks, using a firewall rule listing.

## Fix

- Open5GS: the sources point to a patch on GitHub but do not state a fixed version.

## Timeline

- 2026-11-02T17:24:12+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-106807

## Documents

- [fix] Patch: github.com https://github.com/open5gs/open5gs/commit/d91af3b0ac59f6e9dab25718ef4eed9cbf2813ca

## From the sources

A denial-of-service vulnerability exists in open5GS v2.8.0 in the PFCP transaction handling path. When the PFCP transaction or timer resource pool is exhausted, ogs_pfcp_xact_remote_create() returns NULL, but the failure is not handled properly and leads to an ogs_assert(new) failure in ogs_pfcp_xact_receive(). An attacker able to send PFCP messages to the affected SMF/UPF can exhaust transaction resources and cause the network function process to terminate with SIGABRT, resulting in service disruption and preventing further session establishment until the affected component is restarted.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_