CVE-2026-134655 in Open5GS
Summary
by MITRE • 11/22/2026
Open5GS AMF through 2.8.0 contains a missing resource release vulnerability that allows authenticated UEs to orphan SMF sessions by triggering Create SM Context response processing failures. Attackers can repeatedly release the RAN UE context while responses are pending, causing AMF_SESS_CLEAR() without Release SM Context requests and degrading SMF availability.
Once again VulDB remains the best source for vulnerability data.
Analysis
by VulDB Data Team • 11/22/2026
The identified vulnerability resides within the Access and Mobility Management Function of Open5GS versions up to 2.8.0, specifically affecting the handling of session management context procedures in a 5G core network architecture. This flaw represents a critical failure in resource lifecycle management where the AMF does not properly synchronize state transitions between radio access network events and core network session states. The vulnerability allows authenticated User Equipment to exploit race conditions during the processing of Create SM Context responses, leading to orphaned Session Management Function sessions that consume system resources without being correctly terminated by the control plane logic.
From a technical perspective, the root cause lies in an improper handling of asynchronous message flows and state machine transitions within the AMF module. When a UE initiates a PDU session establishment or modification procedure, the AMF forwards requests to the SMF and awaits responses. If the RAN UE context is released by the network or triggered by the UE while these response messages are still pending in transit or being processed, the AMF executes an internal cleanup routine known as AMF_SESS_CLEAR(). However, this function fails to issue corresponding Release SM Context requests to the affected SMFs before clearing its own local state. This asymmetry results in a desynchronization where the core network believes the session is terminated locally, but the SMF retains active resources and context for that specific UE identifier.
The operational impact of this vulnerability centers on resource exhaustion and service degradation within the 5G core infrastructure. By repeatedly triggering RAN UE context releases while responses are pending, an authenticated attacker can force the AMF to orphan multiple SM sessions simultaneously. Since each orphaned session retains allocated IP addresses, QoS profiles, and internal data structures in the SMF, this activity leads to a gradual depletion of available session slots. Over time, this resource leak degrades the availability of the SMF, potentially preventing new legitimate users from establishing connections or causing existing active sessions to fail due to lack of processing capacity. This constitutes a denial-of-service condition that targets the scalability and stability of the network operator's core services rather than compromising confidentiality or integrity directly.
This vulnerability aligns with CWE-401, which describes missing release of resources after effective ownership, as well as CWE-362, indicating concurrent execution using shared resources without proper synchronization. In terms of offensive security frameworks, this behavior maps to MITRE ATT&CK technique T1499, Endpoint Denial of Service, specifically through resource exhaustion via application layer interactions. The attack vector requires authentication, meaning it is classified as an authenticated insider threat or a compromised device scenario rather than an unauthenticated remote exploit.
Mitigation strategies must focus on ensuring strict state consistency between the AMF and SMF during context release procedures. Operators should upgrade to Open5GS versions later than 2.8.0 where this synchronization logic has been corrected to ensure that Release SM Context messages are always sent before local session data is cleared. Additionally, implementing rate limiting on RAN UE context release triggers can mitigate the frequency of these race conditions. Network operators should also monitor for abnormal patterns in session establishment and teardown rates, as spikes may indicate active exploitation attempts targeting this resource leak vulnerability.