An exploited heap overflow in BIG-IP APM lets an attacker with no login run code on the BIG-IP when APM acts as an OAuth Authorization Server. It affects BIG-IP APM 17.1, 17.5 and 21.1.
Find your fixed versionCVE-2026-94127 is a heap-based buffer overflow (CWE-122) in F5 BIG-IP Access Policy Manager (APM), rated Critical: CVSS v3.1 9.8 and CVSS v4.0 9.3. On affected builds, specially crafted traffic lets an unauthenticated attacker run code when the virtual server has an APM access policy and an OAuth profile configured for the OAuth Authorization Server role. F5 lists 21.1.0.3, 17.5.1.10 and 17.1.3.6 as the first fixed point releases. The table below also lists later fixed releases and engineering hotfixes for earlier builds. If you also need to address vulnerabilities from F5’s separate November 18 notification, use F5’s hotfix selection guidance for that notification. Follow the forensic triage steps below where required.
For BIG-IP APM, F5 identifies exposure when APM is configured as an OAuth Authorization Server on an affected build. F5 says the flaw has been exploited. CISA added it to its Known Exploited Vulnerabilities catalog on September 22, 2026, and listed September 25, 2026, as the due date. Federal civilian agencies must assess each asset’s internet exposure. For this CVE, CISA’s three-day mitigation window and forensic triage also apply to affected internal assets in scope.
F5 advisory K000162605 was published September 22, 2026 and updated October 15, 2026. F5’s release notes confirm the later fixed builds shown here. Find the row for the release you run now.
| Branch | Affected release family before the listed fix | Fixed release | Engineering hotfix |
|---|---|---|---|
| BIG-IP APM 21.x | 21.1.0 | 21.1.0.3 (build 20.0) or 21.1.0.4 (build 3.0) | Hotfix-BIGIP-21.1.0.2.0.30.22-ENG.iso |
| BIG-IP APM 17.x | 17.5.0 through 17.5.1 | 17.5.1.10 (build 35.0) or 17.5.1.11 (build 12.0) | Hotfix-BIGIP-17.5.1.9.0.160.12-ENG.iso |
| 17.1.0 through 17.1.3 | 17.1.3.6 (build 20.0) or 17.1.3.7 (build 4.0) | Hotfix-BIGIP-17.1.3.5.0.41.14-ENG | |
| BIG-IP, all other modules | None | Not applicable | Not vulnerable |
| BIG-IQ Centralized Management | K000162605 marks BIG-IQ as not vulnerable. F5's BIG-IQ release notes list a fix for this CVE. Ask F5 Support to confirm whether your build is affected and which update to apply. | 8.4.2.3 (build 8.0), listed under Fixed in this Release as bug 2524777-9 | — |
| BIG-IP Next CNF and BIG-IP Next for Kubernetes | K000162605 marks both products as not vulnerable. F5 release notes nevertheless list fixes for this CVE. Ask F5 Support to confirm exposure and the appropriate update for your platform and build. | F5 lists this CVE under Fixed Issues for CNF 2.2.5 on AON, OpenShift and Robin, and for BIG-IP Next for Kubernetes 2.2.5. The CNF 2.4.1 release notes for AON, OpenShift and Robin also list this CVE as fixed. | — |
| F5OS, F5OS-A, F5OS-C | None | Not applicable | Not vulnerable |
| NGINX, F5 Distributed Cloud, NGINX One Console, F5 AI Gateway | None | Not applicable | Not vulnerable |
/shared/images. If you already run an engineering hotfix, check that the target base image or hotfix includes the fixes you rely on. Open a case with F5 Support if any are missing or unlisted. F5’s advisory writes the 17.1 hotfix name with a space (“Hotfix-BIGIP 17.1.3.5.0.41.14-ENG”); F5’s CVE record uses a hyphen, as in the table. F5 evaluates only releases that have not reached End of Technical Support. If you run a branch that is not in the table, treat it as unassessed and move to a fixed release. F5’s CVE record starts the 17.1 range at 17.1.0. NVD’s CPE analysis starts it at 17.0.0. If you run 17.0.x with APM configured as an OAuth Authorization Server, check your build with F5 Support and plan to move to a fixed release.
The flaw is in the OAuth Authorization Server feature of APM, on the traffic side of the BIG-IP.
F5 describes it this way: “When a BIG-IP APM access policy and an OAuth profile are configured on a virtual server, specific malicious traffic can lead to remote code execution (RCE).” F5 tracks it as bug ID 2524777 and classifies it as CWE-122, heap-based buffer overflow. F5 found it internally.
The impact, in F5’s words: it “allows an unauthenticated attacker to perform RCE.” The CVSS vectors from F5 (v3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H) mean network reachable, low complexity, no privileges and no user interaction. CISA’s enrichment of the CVE record marks exploitation as active and the flaw as automatable.
Three details narrow who is exposed:
Use F5’s BIG-IP documentation to check the running software and identify virtual servers configured as OAuth Authorization Servers.
F5’s K8759: in the Configuration utility, go to System > Configuration; the version is in the Version box. From the command line, enter tmsh, then:
show /sys version
Compare the running version with the table above. In tmsh, use show /sys version detail to verify applied hotfixes. Boot into the volume containing the fixed release or hotfix, then verify the running version on every BIG-IP that runs APM, including both units of an HA pair.
Installing a fixed release or the engineering hotfix removes the flaw. The steps below only reduce the risk.
F5’s only listed mitigation is an iRule applied to the affected BIG-IP APM virtual server. It is not published in the advisory: open a case with F5 Support to get it. CISA’s catalog entry calls the iRule a temporary mitigation that allows time for forensic triage, and says to install the final vendor patch as soon as possible after that.
Use the mitigation iRule supplied by F5 Support.
This is general practice, not an F5 workaround. If an OAuth authorization server serves only internal users or a known set of partner systems, limit its virtual server to those source networks on the BIG-IP or on the firewall in front. Anything that can still reach it can still attempt the exploit.
CISA flags this CVE for forensic triage under BOD 26-04. For affected federal civilian assets in scope, apply the F5 iRule as temporary mitigation and follow CISA’s forensic triage guidance. Collect volatile data and preserve relevant logs and core files before patching or rebooting when possible, even if an initial log review shows no suspicious activity. Install the final vendor patch as soon as triage is complete. CISA directs agencies to follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. The checks below can help identify signs of prior compromise.
F5 publishes a pattern to look for, not a list of attacker addresses or files.
F5 says three indicators seen close together in time are a reason to investigate: multiple OAuth authentication failures, followed by suspicious commands, shortly followed by a TMM SIGABRT. Any one of them alone does not mean compromise; the combination and frequency have been seen to correlate with an attack.
F5 rates this a medium-confidence indicator. The message is normal in some situations, so the signal is repetition: 10 or more in one log, especially all from a single IP address, should get a human review. The message F5 quotes:
<DATE> <HOST> err tmm1[30975]: 01990004:3: <PROFILE_NAME>: Request UserInfo from Source ID (null) IP <IP> failed. Error Code (invalid_token) Error Description (The access token is invalid.)
F5 also gives this command for APM OAuth statistics. An unexplained increase in total_failed should be investigated:
tmctl global_oauth_stat -s total_requests,total_userinfo_requests,total_failed
If you see the OAuth failures, especially in rapid succession or large volume, note the timestamps and review /var/log/audit around those timestamps for suspicious commands, including commands attributed to legitimate accounts.
F5 has observed TMM enter a loop, after which the SOD daemon sends it a SIGABRT. A TMM core file alone is not an indicator, but F5 says core files should be investigated.
Check for unauthorized changes to the TMM hook that runs after a crash. watchTowr Labs demonstrated a change to that hook in its lab.
| Date | Event |
|---|---|
| September 22, 2026 | F5 publishes advisory K000162605 and says the flaw has been exploited. The CVE record is published. CISA adds CVE-2026-94127 to the Known Exploited Vulnerabilities catalog. |
| September 25, 2026 | Due date listed in CISA KEV. BOD 26-04 governs the applicable actions and timing for each federal civilian agency asset. |
| October 15, 2026 | Latest update to F5 advisory K000162605 as of this alert. It lists fixed releases 21.1.0.3, 17.5.1.10 and 17.1.3.6 and the engineering hotfixes. |
| October 23, 2026 | F5 releases BIG-IP 21.1.0.4, 17.5.1.11 and 17.1.3.7. The release notes for all three confirm they include the fix for CVE-2026-94127. |
A known-attacker list can cut down who reaches your edge. It does not fix this flaw.
CVE-2026-94127 is a software flaw, and only F5’s fixed release or hotfix removes it. A block list does not stop the exploit: an attacker using an address that is not on the list can still send the malicious traffic.
What a block list can do is reduce the attack traffic that reaches the services you publish. ThreatListPro covers 5.4M+ IPv4 addresses, including addresses covered by CIDR ranges. The list is built from real attacks on live production Palo Alto GlobalProtect VPN portals in the US and is checked for new attackers every 6 hours. ThreatListPro has no BIG-IP setup guide; it is loaded on the firewall in front of the BIG-IP, as a deny rule for inbound traffic. The setup instructions cover the supported firewalls. The list is IPv4 only.
Yes. F5 says this vulnerability has been exploited. F5 published its advisory and CISA added CVE-2026-94127 to KEV on September 22, 2026. The catalog lists September 25, 2026, as the due date. CISA’s three-day mitigation window and forensic triage also apply to affected internal assets in scope for federal civilian agencies.
F5 lists BIG-IP 21.1.0.3, 17.5.1.10 and 17.1.3.6 as the first fixed point releases. It also lists three engineering hotfixes: Hotfix-BIGIP-21.1.0.2.0.30.22-ENG, Hotfix-BIGIP-17.5.1.9.0.160.12-ENG and Hotfix-BIGIP-17.1.3.5.0.41.14-ENG. The later releases in the table also include the fix. F5 identifies affected builds in the 21.1.0, 17.5.0 through 17.5.1, and 17.1.0 through 17.1.3 release families when APM is configured as an OAuth Authorization Server.
No, according to F5. The flaw is only present when BIG-IP APM is configured as an OAuth Authorization Server, with an access policy and an OAuth profile on a virtual server. F5 says deployments that use APM strictly as an OAuth Client or Resource Server, without OAuth authorization server profiles, are not affected.
No. F5 says the flaw allows an unauthenticated attacker to perform remote code execution. The CVSS v3.1 score is 9.8 and the CVSS v4.0 score is 9.3, both Critical, with no privileges or user interaction required.
Open a case with F5 Support to request the iRule and apply it to the affected BIG-IP APM virtual server. For assets subject to CISA’s forensic triage requirement, use the iRule as a temporary mitigation, complete triage, and install the final vendor patch as soon as possible.
F5 calls this a data plane issue with no control plane exposure. The risk is the virtual server that carries the APM access policy and the OAuth profile, not the management port. F5 also says BIG-IP systems in Appliance mode are vulnerable.
F5 lists no IP addresses or file hashes. Investigate repeated OAuth failures in /var/log/apm followed by suspicious commands in /var/log/audit and then a TMM SIGABRT shortly afterward. On its own, 10 or more of those OAuth failures in one log already warrants a human review, and F5 says any TMM core file should be investigated.
ThreatListPro covers 5.4M+ IPv4 addresses and is checked for new attackers every 6 hours. Monthly is $19.99 with a 14-day free trial and no card needed.
Start a 14-day free trial Or test the free sample first: no signupThreatListPro uses a plain-text feed for firewalls that support external IPv4 address lists.
More dated alerts, and how to block addresses at the firewall in front of an exposed service.