DEV Community

Cover image for 6 Critical F5 BIG-IP Flaws: Affected Versions and Fixes
devTalk
devTalk

Posted on Originally published at dev-talk.com

6 Critical F5 BIG-IP Flaws: Affected Versions and Fixes

Last updated: 22 November 2026. F5 disclosed these flaws on 18 November 2026 and revised some advisories on 19 November, so details may have changed since. I could not open F5's own support articles from my environment, so the facts below come from security vendors, government advisories and news reports that cite F5. SeeHow this was checked.

On 18 November 2026, F5 published advisories for six critical vulnerabilities in the data plane of BIG-IP, the part that handles live traffic. All six are stack-based buffer overflows. An unauthenticated remote attacker can crash the Traffic Management Microkernel (TMM), which interrupts traffic until it restarts, and F5 says the flaws may also allow remote code execution. F5 lists no workaround. The fix is an engineering hotfix. Each flaw needs a specific feature on a virtual server, which means you can check whether you are exposed without guessing.

The short version

  • Six CVEs, all rated 9.8 (CVSS 3.1) and 9.3 (CVSS 4.0).
  • No authentication needed, but each needs a trigger configuration on a virtual server.
  • The management interface is not the attack path, so locking it down does not help.
  • No exploitation has been reported by F5 at the time of the sources I read, and the CVEs were not in CISA's Known Exploited Vulnerabilities catalog when one vendor checked.
  • Fixes are engineering hotfixes only. No scheduled release containing them was listed.

The six flaws and what triggers each

CVE F5 article Trigger on a virtual server Product
CVE-2026-123637 K000163937 APM per-request policy containing both the Client Information and Endpoint Security agents BIG-IP APM
CVE-2026-121805 K000163946 iRule using PEM::subscriber on client traffic BIG-IP PEM, BIG-IP Next for Kubernetes, CNF
CVE-2026-120292 K000163940 HTTP profile plus an iRule using MR::message route All BIG-IP modules, BIG-IP Next for Kubernetes
CVE-2026-120285 K000163939 HTTP profile plus an iRule using PSC::ip_address All BIG-IP modules, BIG-IP Next for Kubernetes, CNF
CVE-2026-116953 K000163941 iRule using GENERICMESSAGE::route on client traffic All BIG-IP modules, BIG-IP Next for Kubernetes
CVE-2026-120722 K000163945 iRule using DNS::scrape on client traffic BIG-IP DNS, BIG-IP Next for Kubernetes

Five of the six are iRule commands. A device that does not use these commands, and does not run the APM policy combination, does not meet the trigger. That is the most useful fact in the whole advisory, and it also means the answer is in your own configuration.

A note on the count: most sources, including H-ISAC and the UK NHS alert, say six critical flaws. BleepingComputer's article names five (it leaves out DNS::scrape, CVE-2026-120722) and describes a sixth, CVE-2026-126822, as a separate high-severity APM SAML flaw. The Canadian Cyber Centre lists K articles for all six critical CVEs above. I treat six as the count.

Affected products and fixed builds

The Canadian Cyber Centre advisory lists affected versions as "prior to" the fixed build:

Branch Fixed build
BIG-IP 21.1.0 BIGIP-21.1.0.4.0.19.3-ENG
BIG-IP 17.5.0 to 17.5.1 BIGIP-17.5.1.11.0.61.12-ENG (or BIGIP-17.5.1.10.0.80.35-ENG on specific starting builds)
BIG-IP 17.1.0 to 17.1.3 BIGIP-17.1.3.7.0.21.4-ENG (or BIGIP-17.1.3.6.0.14.20-ENG on a specific starting build)
BIG-IP Next for Kubernetes 2.2 / 2.3 / 2.4 BNK-2.2.5-EHF1, BNK-2.3.4-EHF1, BNK-2.4.2-EHF1
BIG-IP CNF 2.2 / 2.3 / 2.4 CNF-2.2.5-EHF1, CNF-2.3.4-EHF1, CNF-2.4.2-EHF1

Which hotfix build you get depends on your current build, and F5 documents that mapping in K000164239. Do not pick a build from this table alone. BIG-IQ, F5OS, F5 Distributed Cloud, NGINX and F5 AI Gateway are reported as not affected. F5 does not evaluate versions at End of Technical Support, so treat those as exposed if a trigger is configured. If you run a custom engineering hotfix today, you need a new one from F5 Support.

F5 also says engineering hotfixes do not get the extensive QA of scheduled releases and carry no warranty. Plan a maintenance window with a rollback, test on a standby, and do not assume a hotfix is a drop-in.

Check whether you are exposed

The steps below follow the read-only checks described by Rescana, which is a security vendor and not F5. Run them on the device with an account that can read the configuration.

  1. Check the version and provisioned modules:
tmsh show sys version
tmsh list sys provision one-line
Enter fullscreen mode Exit fullscreen mode
  1. Export every iRule and search for the five commands:
tmsh -q -c 'cd /; list ltm rule recursive' > /var/tmp/irules.txt
Enter fullscreen mode Exit fullscreen mode

I wrote a small script for the search and ran it against a sample file with one rule per command, plus near-miss lines. It matched all five commands (and a commented-out mention) and ignored MR::message_foo and DNS::scrapeX:

#!/bin/sh
# Usage: f5-irule-scan.sh irules.txt
# A hit means "review this rule", not "this device is exploitable".
f="${1:?usage: $0 irules.txt}"
grep -nE '(PEM::subscriber|MR::message[[:space:]]+route|PSC::ip_address|GENERICMESSAGE::route|DNS::scrape)([^A-Za-z0-9_:]|$)' "$f"
Enter fullscreen mode Exit fullscreen mode
3:        set s [PEM::subscriber config ]
8:        MR::message route
13:        log local0. "ip [PSC::ip_address]"
18:        GENERICMESSAGE::route "pool1"
23:        DNS::scrape
28:        # old: PEM::subscriber was removed
Enter fullscreen mode Exit fullscreen mode

The script exits 1 when it finds nothing. That is not a clean bill of health on its own: rules created by automation may live in /config/bigip.conf or /config/partitions/*/bigip.conf, and the APM trigger is a policy, not an iRule. I tested the script on sample text, not on a real BIG-IP export.

  1. Find out which virtual servers use the matching rules, and whether the HTTP profile is on the same virtual server for the two HTTP-profile flaws:
tmsh -q -c 'cd /; list ltm virtual recursive rules profiles destination'
Enter fullscreen mode Exit fullscreen mode
  1. For APM, open the Visual Policy Editor and look for a per-request policy that contains both the Client Information and Endpoint Security agents, then list which virtual servers use per-request policies:
tmsh -q -c 'cd /; list ltm virtual recursive per-flow-request-access-policy'
Enter fullscreen mode Exit fullscreen mode
  1. On Kubernetes, export your iRule resources from Git or with kubectl and run the same search, and record the BNK or CNF version for each cluster.

I did not run any of the tmsh commands, because I have no BIG-IP to test on. Check the syntax on your version before using them in a runbook.

What to do first

  1. Rank devices by exposure. A matching virtual server that faces the internet or a partner network comes first.
  2. Get the hotfix build for your branch and current build from F5, using K000164239.
  3. Patch the standby of each high-availability pair, fail over, verify with tmsh show sys version, then patch the former active unit.
  4. If you cannot patch yet, the only precondition-removing options are to detach or rewrite the matching iRule or to restrict source addresses to the affected virtual server. F5 lists no mitigation, so this is vendor-neutral advice from Rescana. Test it first, because those iRule commands usually do real work, and removing an APM posture check weakens a security control.
  5. Make sure failover works. HA reduces the impact of a TMM crash, but the peer runs the same configuration, so it does not remove the flaw.

How to tell if TMM is crashing

No indicators of compromise have been published. These are monitoring leads from Rescana:

  • TMM restarts. Core files go to /shared/core, and logs are in /var/log/tmm and /var/log/ltm. Alert on any restart and on unexpected failovers.
  • Repeated restarts tied to one client source or one virtual server deserve a call to F5 Support.
  • After a crash, review configuration change logs, /var/log/audit, and any new or altered iRules.
  • If you suspect compromise, collect a QKView and the core files before rebooting, then rotate TLS keys, certificates and stored credentials.

Because the possible impact includes code execution, a crash is worth treating as an incident until you know why it happened.

Related flaws in the same release

F5 disclosed more than the six critical issues that day. Three more APM TMM crashes (CVE-2026-125272, CVE-2026-126822 and CVE-2026-121812) are fixed by the same hotfixes. Three others are fixed in scheduled releases 21.1.0.3, 17.5.1.10 and 17.1.3.6: CVE-2026-126336, CVE-2026-125274 (iControl SOAP, 8.8) and CVE-2026-120718 (iControl REST token expiry, 8.1). For the iControl SOAP issue, F5's listed mitigation is limiting access to trusted networks, and that can break high-availability setups if you block self-IPs.

A different APM flaw, CVE-2026-94127, was disclosed on 22 September 2026 and exploited. It is a separate heap-based issue with a different trigger and is not one of these six. If you did not patch that one, start there.

Frequently asked questions

Can attackers exploit these without logging in?

Yes, per F5, but only against a virtual server that has the trigger feature configured.

Does restricting the management interface help?

No. These are data-plane issues. F5 says the management interface is not exposed. Restricting who can reach the affected virtual server does.

Is there a workaround?

F5 lists none for the six critical flaws. Only engineering hotfixes are available.

Are these flaws being exploited?

F5 has not said so in the sources I read, and one vendor reported no public exploit code and no KEV listing as of 18 November. That can change quickly. Check CISA's catalog and F5's advisories again before you set your patch timeline.

Which products are not affected?

BIG-IQ, F5OS, F5 Distributed Cloud, NGINX products and F5 AI Gateway are reported as not affected by these six.

Related Dev Talk articles

update log

  • 22 November 2026: first version.
  • To recheck on each update: F5's advisories (K000164058 and K000164239), any change to the affected product lists, CISA KEV status, reports of exploitation, and a scheduled release that includes the fixes.

Sources


Originally published on DEV Talk.

Top comments (0)