
F5 warns of six critical vulnerabilities in BIG‑IP that can cause TMM crash and enable RCE
F5 disclosed six critical flaws in BIG‑IP that allow crashing the Traffic Management Microkernel (TMM) and, in some scenarios, executing remote code. The company released engineering hotfixes without guaranteed functionality.
Six critical vulnerabilities in BIG‑IP can interrupt traffic by causing failures in the management microkernel, as well as open the door to remote code execution.
F5 Networks announced today (November 18, 2026) the discovery of six critical vulnerabilities in its BIG‑IP platform. The flaws, scored 9.8 (CVSS v3.1) and 9.3 (CVSS v4.0), are stack-based buffer overflows that can cause a crash of the Traffic Management Microkernel (TMM) without authentication, interrupting traffic until the microkernel is restarted. In some of the described scenarios, the attacker can also achieve remote code execution (RCE).
The vulnerabilities require specific configurations on virtual servers: five of them depend on iRules that use particular commands, while the sixth requires a per‑request policy in the APM module that simultaneously includes the Client Information and Endpoint Security agents. The flaws affect various BIG‑IP variants, including BIG‑IP Next for Kubernetes and BIG‑IP Cloud‑Native Edition (CNF), although the latter is not vulnerable to one of the CVEs according to the official registry.
The identifiers of the vulnerabilities are:
- CVE‑2026‑123637 – APM policy with Client Information and Endpoint Security agents.
- CVE‑2026‑121805 – iRule that uses
PEM::subscriberon client traffic. - CVE‑2026‑120292 – HTTP profile and iRule that uses
MR::message route. - CVE‑2026‑120285 – HTTP profile and iRule that uses
PSC::ip_address. - CVE‑2026‑116953 – iRule that uses
GENERICMESSAGE::route. - CVE‑2026‑120722 – iRule that uses
DNS::scrape.
To mitigate the risks, F5 released engineering hotfixes (EHF) specific to each affected variant. The patched versions include BNK‑2.4.2‑EHF1, BNK‑2.3.4‑EHF1 and BNK‑2.2.5‑EHF1 for BIG‑IP Next for Kubernetes, as well as CNF‑2.4.2‑EHF1, CNF‑2.3.4‑EHF1 and CNF‑2.2.5‑EHF1 for BIG‑IP CNF for CVEs 2026‑121805 and 2026‑120285. The company warns that these hotfixes do not undergo the same quality control rigor as scheduled releases and are provided without efficacy guarantees.
In addition to the six critical flaws, the notification includes six other lower‑severity vulnerabilities (CVSS v3.1 between 7.5 and 8.8) that can also be exploited in specific scenarios, such as iControl SOAP, iControl REST, and various APM configurations. Each has fixes included in the same engineering hotfixes or in specific maintenance releases.
So far, F5 has not received reports of active exploitation of the critical vulnerabilities. However, the company notes that four of the flaws were identified by an external researcher (Sina Kheirkhah, WatchTowr) and the rest were discovered internally. The lack of known exploitation does not eliminate the need for immediate application of the hotfixes, given the severity of the CVSS scores and the potential for total traffic disruption.
Organizations using BIG‑IP should immediately review their iRule and APM policy configurations, disable unnecessary agents, and verify that applicable hotfixes have already been installed. Monitoring unexpected TMM restarts can serve as an early indicator of an exploitation attempt.
CyberX Technical Analysis
The six vulnerabilities disclosed by F5 are all stack-based buffer overflows that occur in the data plane of BIG‑IP. The exploitation mechanism depends on specific commands inserted into iRules or APM policies that combine information‑collection agents. When the vulnerable command is processed, the overflow corrupts the stack of the Traffic Management Microkernel (TMM), causing an immediate crash of the process responsible for forwarding traffic.
The TMM is the central component that manages real‑time packet flow. Its termination forces an automatic restart, temporarily interrupting all traffic passing through the device. In high‑availability environments, the failure can trigger unexpected failover, overloading redundant devices and potentially causing packet loss.
Although F5 indicates that the flaws may allow remote code execution (RCE), full exploitation requires the attacker to inject arbitrary code after the overflow, which depends on the presence of vulnerable libraries or functions in the TMM context. So far, there is no public evidence of successful exploitation that achieves RCE.
The vulnerabilities affect multiple versions and variants of BIG‑IP, including Kubernetes deployments (BIG‑IP Next) and the Cloud‑Native edition (CNF). The versions that still receive technical support are those that receive engineering hotfixes. Versions outside the support cycle have not been confirmed as secure, implying residual risk for customers still operating on legacy versions.
Technical Recommendations:
- Apply the corresponding engineering hotfixes (EHF) for the affected versions immediately, according to the fix matrix published by F5.
- Review all iRules in use and remove or replace those that use the vulnerable commands (
PEM::subscriber,MR::message route,PSC::ip_address,GENERICMESSAGE::route,DNS::scrape). - Audit APM policies to ensure that there are no combinations of Client Information and Endpoint Security agents in per‑request requests unless strictly necessary.
- Implement monitoring of TMM restart events and service failure alerts to quickly detect potential exploitation attempts.
- Consider migrating to BIG‑IP versions that are already out of the hotfix cycle but receive regular security updates.
- Apply restricted access controls to management interfaces and limit exposure of control APIs to trusted networks.
These measures reduce the attack surface and mitigate the impact if an attacker manages to trigger one of the described vectors.
CyberX SOC Intelligence Bulletin
F5 disclosed six critical vulnerabilities in BIG‑IP that can cause a crash of the Traffic Management Microkernel (TMM) without authentication and, in some scenarios, enable remote code execution. So far, there are no confirmed indicators of compromise (IoCs) in public sources.
Indicators of Compromise
No binary hashes, IP addresses, domains, or malware signatures associated with active exploitation of these CVEs were identified in the consulted sources.
Detection
- Monitor BIG‑IP logs for error messages related to TMM, such as "TMM crash" or "TMM restart".
- Correlate unexpected reboot events with the presence of iRules containing the listed vulnerable commands.
- Use front‑end service availability metrics to detect sudden outages that coincide with the execution of specific APM policies.
Mitigation
- Apply the engineering hotfixes (EHF) provided by F5 for the affected versions.
- Disable or remove iRules that use the vulnerable commands until the hotfix is in production.
- Review and simplify APM policies, avoiding the combination of Client Information and Endpoint Security agents in per‑request submissions.
- Implement restricted access controls to the BIG‑IP management interfaces.
Response
- In the event of a TMM crash, isolate the affected device, collect diagnostic logs, and open a support ticket with F5 to obtain additional hotfixes.
- Preserve evidence of reboot events and iRules/APM configuration for forensic analysis.
- Notify the security and network operations teams about the need to review traffic policies.
Frequently Asked Questions
What are the critical vulnerabilities announced by F5 in BIG-IP?
F5 announced six critical stack-based buffer overflow vulnerabilities scored 9.8 and 9.3 that can crash the Traffic Management Microkernel (TMM) without authentication and potentially enable remote code execution.
What causes the TMM crash in these vulnerabilities?
The vulnerabilities occur in the data plane and are triggered when specific commands inserted into iRules or APM policies process, corrupting the stack of the Traffic Management Microkernel and causing an immediate crash.
Are there available patches or hotfixes?
Yes, F5 released specific engineering hotfixes (EHF) for affected variants, such as BIG-IP Next for Kubernetes and BIG-IP Cloud-Native Edition, though the company warns they do not undergo standard quality control.
Have these vulnerabilities been actively exploited in the wild?
So far, F5 has not received reports of active exploitation or identified indicators of compromise associated with these critical vulnerabilities.


