Automated Threat Modeling

Brief Summary: F5 BIG-IP APM CVE-2026-123637, a Stack-Based Buffer Overflow in TMM Enabling DoS and Possible RCE

A brief summary of CVE-2026-123637, a CWE-121 stack-based buffer overflow in F5 BIG-IP APM that lets remote unauthenticated requests crash TMM and may lead to code execution. Includes affected versions and F5's guidance on choosing an engineering hotfix build.

CVE Analysis

7 min read

Gaurav Sarraf
Gaurav Sarraf

2026-11-18

Brief Summary: F5 BIG-IP APM CVE-2026-123637, a Stack-Based Buffer Overflow in TMM Enabling DoS and Possible RCE

Introduction

A stack-based buffer overflow in F5 BIG-IP Access Policy Manager (APM) allows remote, unauthenticated requests to crash the Traffic Management Microkernel (TMM), interrupting all traffic the device handles until the process restarts. Because the overflow occurs in TMM, the BIG-IP data-plane process, F5 also warns that the condition may lead to remote code execution. The issue carries a 9.8 Critical rating under CVSS 3.1 and 9.3 Critical under CVSS 4.0.

The exposure is narrower than the score alone suggests. It requires APM to be provisioned and a specific per-request policy configuration on a virtual server. F5 has shipped fixes only as engineering hotfixes and lists no mitigation, so for most teams the main operational question is which build to install.

Technical information

Root cause

F5 classifies CVE-2026-123637 as CWE-121: Stack-based Buffer Overflow in TMM. The vulnerable code path is reachable only when both of the following are true:

  • BIG-IP APM is provisioned.
  • A per-request policy that includes the Client Information agent and Endpoint Security agents is applied to a virtual server.

F5 describes the trigger only as "undisclosed requests." The advisory does not name the vulnerable function, the field or parameter involved, or a protocol sequence. BIG-IP is closed source, so there is no public commit or diff to examine.

We can say with reasonable confidence that a stack overflow inside a data-plane process explains both outcomes F5 lists. Overwriting stack memory in TMM can corrupt state and crash the process, producing a denial of service. Depending on what is overwritten (for example saved return data), it may be possible to redirect execution. F5's wording is that the issue "may lead to" RCE, and the advisory does not provide technical details demonstrating code execution.

Our reading of the fix, based on the CWE classification, is that the patched TMM bounds or validates attacker-supplied data handled by the Client Information and Endpoint Security agents before copying it into fixed-size stack buffers. F5 has not published the code change, so treat this as an inference rather than a confirmed detail.

Scoring

The vectors published for this CVE are:

CVSS 3.1: AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H (9.8 Critical) CVSS 4.0: AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N (9.3 Critical)

The two scores differ only because they use different CVSS versions. Both describe the same impact. When recording severity internally, note which version you are using.

Attack flow

Based on what F5 has disclosed, the attack path looks like this:

  1. The attacker identifies a BIG-IP virtual server with APM provisioned and a per-request policy that combines the Client Information agent with Endpoint Security agents.
  2. Without authentication or user interaction, the attacker sends crafted requests (contents undisclosed by F5) to that virtual server.
  3. The agents process attacker-controlled data, which overflows a stack buffer in TMM.
  4. TMM terminates. Traffic through the system is disrupted while TMM restarts.
  5. Under conditions F5 has not described, the overflow may be leveraged for code execution in the context of TMM.

F5 states there is no control-plane exposure. This is a data-plane issue only. That limits the reachable surface to traffic arriving at the affected virtual server, but those virtual servers are, by design, the internet-facing access points that APM protects.

Patch information

F5 tracks the issue internally as bug ID 2567577. The fix is available only through engineering hotfixes (EHFs). No scheduled maintenance or point release contains it yet. Our understanding of the fix comes from advisory K000163937 and the build-selection article K000164239.

Fixed builds

  • 21.x (vulnerable: 21.1.0): BIGIP-21.1.0.4.0.19.3-ENG
  • 17.5.x (vulnerable: 17.5.0 through 17.5.1): BIGIP-17.5.1.11.0.61.12-ENG or BIGIP-17.5.1.10.0.80.35-ENG
  • 17.1.x (vulnerable: 17.1.0 through 17.1.3): BIGIP-17.1.3.7.0.21.4-ENG or BIGIP-17.1.3.6.0.14.20-ENG

The build names follow F5's pattern of base version, then EHF build number, with the ENG suffix marking an engineering hotfix. Builds are available from MyF5 Downloads.

Choosing the right build

Some branches have two fixed builds because F5 rebuilt hotfixes for customers already running earlier EHFs. K000164239 maps current versions to the correct build:

  • 17.1.3.7 or any earlier base release: BIGIP-17.1.3.7.0.21.4-ENG
  • Already on BIGIP-17.1.3.6.0.9.20-ENG: BIGIP-17.1.3.6.0.14.20-ENG
  • 17.5.1.11 or any earlier base release: BIGIP-17.5.1.11.0.61.12-ENG
  • Already on BIGIP-17.5.1.10.0.7.35-ENG or BIG-IP-17.5.1.10.0.37.35-ENG: BIGIP-17.5.1.10.0.80.35-ENG
  • 21.1.0.4 or any earlier base release: BIGIP-21.1.0.4.0.19.3-ENG
  • Any other custom EHF: ask F5 Support to build a new hotfix that includes the fix.

Environments that already run point-fix EHFs should follow this mapping closely. Moving to the generic base-release build could drop fixes you depend on, which is why F5 produced separate rebased hotfixes.

Limits of an engineering hotfix

F5 notes that EHFs "do not undergo the extensive QA assessment of scheduled software releases" and come with no warranty of usability. We recommend testing in staging first and keeping a rollback image ready on another boot location. F5 is expected to fold the fix into a future scheduled release. Under F5's versioning policy, once a fix lands in a branch, every later release in that branch includes it. F5 lists no mitigation, so installing the hotfix is the only remediation it offers. If no fix candidate is listed for your branch, F5 recommends upgrading to a version with the fix.

Scope of the fix

Only the BIG-IP APM module needs a hotfix. F5 lists every other BIG-IP module, BIG-IQ, F5OS, NGINX, BIG-IP CNF and BIG-IP Next for Kubernetes as not vulnerable. One upstream summary claimed a BIG-IP Next for Kubernetes hotfix (BNK-2.4.2-EHF1) applies to this CVE. That is incorrect: the advisory states BIG-IP Next for Kubernetes is not affected by CVE-2026-123637.

Affected systems and versions

Affected product: F5 BIG-IP APM, only in the following configuration:

  • APM is provisioned, and
  • A per-request policy including the Client Information agent and Endpoint Security agents is applied to a virtual server.

Affected releases listed by F5:

  • 21.x: 21.1.0
  • 17.5.x: 17.5.0 through 17.5.1
  • 17.1.x: 17.1.0 through 17.1.3

Not vulnerable per F5: all other BIG-IP modules, BIG-IQ, F5OS, NGINX, BIG-IP CNF, and BIG-IP Next for Kubernetes.

F5 did not evaluate software versions that have reached End of Technical Support (EoTS). If an EoTS release is missing from the list above, that does not mean it is unaffected.

Vendor security history

F5 is the CNA for this CVE and credits Sina Kheirkhah (SinSinology) of watchTowr with reporting it. CVE-2026-123637 was published alongside other BIG-IP issues in F5's November 18, 2026 out-of-band security notification, which contemporaneous coverage described as a batch of critical BIG-IP flaws that can crash traffic in specific setups.

As broader context only: F5 disclosed a security incident in October 2025 involving a threat actor's persistent access to F5 systems and exfiltration of certain files. CISA's Emergency Directive 26-01 described a nation-state-affiliated actor's compromise of F5 systems, including exfiltration of BIG-IP source code and vulnerability information. None of the sources connect that incident or actor to CVE-2026-123637.

On exploitation status, a November 2026 report states F5 reported no exploitation and that the CVE was not in the CISA Known Exploited Vulnerabilities catalog at the time. We found no public proof of concept or attributed activity. That reflects status at disclosure and does not rule out later exploitation.

References

Detect & fix
what others miss

Works with
  • GitHub
  • GitLab
  • Bitbucket
  • Azure DevOps Services
  • Gerrit
  • CVS
  • Jira
  • Linear
  • Slack
  • Security Compass
Explore integrations
Security magnifying glass visualization