IPS virtual patching: signature and filter checks
An IPS signature counts as a mitigation when it blocks, sits inline and is current. For CVE-2026-0322, Palo Alto named Threat IDs 510046 and 510047 on 14 Oct 2026, calls the coverage limited, and requires inbound decryption to make them work. Check the action, the path and the content version.
On this page
What a signature covers
An intrusion prevention system inspects traffic and drops flows that match a signature written for one vulnerability. OWASP's Virtual Patching Cheat Sheet names the device class: "Intermediary devices such as a WAF or IPS appliance". NIST SP 800-53 SI-4 places it in monitoring tooling: "System monitoring capabilities are achieved through a variety of tools and techniques, including intrusion detection and prevention systems, malicious code protection software, scanning tools, audit record monitoring software, and network monitoring software."
A WAF rule reads application-layer requests for one web application. A signature sits on the network path, so it can cover many hosts and non-web protocols at once. The weakness is the same: the sensor has to see the traffic in readable form. Signatures are written against known behavior, so a variant can slip past one, and a flaw with no published signature has no protection.
Worked case: PAN-OS management interface, CVE-2026-0322
On 14 Oct 2026 Palo Alto Networks published its advisory for CVE-2026-0322, a code injection flaw (CWE-94, CVSS 4.0 base score 8.6) in the PAN-OS management interface. An attacker who can reach the interface and holds a valid high-privilege account can run code as root. CISA added it to KEV the same day with a due date of 17 Oct 2026, and Palo Alto's exploitation status reads "Very limited exploitation has been observed targeting the Palo Alto Networks management interface." Every supported train has a fixed build; the per-train fixed builds are in the table on the access restriction page, and the table below compares the two options Palo Alto gives.
Palo Alto states that no workaround exists, then names a signature: Threat Prevention subscribers can enable Threat IDs 510046 and 510047 on Applications and Threats content version 9159-10287 or later. Palo Alto calls this limited coverage, and it works only if all four of these hold:
- management-port traffic is routed through a data-plane interface, using a management profile on that interface
- the certificate for inbound traffic management is replaced
- inbound traffic to the management interface is decrypted
- threat prevention is enabled on that traffic
The cost, per the same record, is SSL decryption load and certificate work on the firewall, and attack variants may still pass. What can be checked on your own sensor is the content version it runs: it must be 9159-10287 or later.
| Option | What it needs | What stays open |
|---|---|---|
| Threat IDs 510046 and 510047 | The four conditions above and content 9159-10287 or later | Variants that do not match; the code remains |
| Restrict the management interface to one jump box | See access restriction | Anyone with a high-privilege account who can reach the interface from the jump box; the code remains. Palo Alto rates the vector 8.5 instead of 8.6 |
Three checks before you rely on it
- Prevent, not detect. A signature set to alert logs the attack and lets it through. On PAN-OS, go to Objects > Security Profiles > Vulnerability Protection, open the profile the security rule uses, select the Exceptions tab and click Show all signatures. Filter for 510046 and 510047 and read the Action column: it must block (Palo Alto's documentation recommends reset-both for critical, high and medium severity) and not alert. Palo Alto's page notes that an exception can change the response to one signature, so look for an exception on these two IDs.
- Placement. Confirm the sensor is inline on the path and that decryption happens upstream of it. For CVE-2026-0322 this is the whole control: without decryption of management traffic the signature sees nothing.
- Freshness. Under Device > Dynamic Updates, find the Applications and Threats row with a check mark in the Currently Installed column. Its version must be 9159-10287 or later; anything older does not carry the Threat IDs.
Then apply the two SP 800-40 Rev. 4 checks to the result: the control is "functioning as intended" (confirm a block log entry for the signature, not only that it is loaded), and the response "continues to be in place", so re-check after every content update, route change and policy push.
What a block looks like in the log: in Monitor > Logs > Threat, an entry for these signatures has type THREAT and subtype vulnerability, the Threat ID 510046 or 510047, and an Action that blocks. Palo Alto's threat log field list defines the values: "alert" is "threat or URL detected but not blocked", while drop, reset-client, reset-server, reset-both and block-ip are blocking actions (reset-both sends a TCP RST to both client and server). An entry with Action alert means the signature fired and let the traffic through. Filter by the Threat ID column, not by name (Palo Alto Networks, threat log fields, read 19 Nov 2026).
Second case: a FortiOS virtual patch
Fortinet advisory FG-IR-26-273 (CVE-2026-126259, published 10 Nov 2026, updated 16 Nov 2026) describes a missing authentication flaw [CWE-306] in FortiOS that may allow an unauthenticated attacker with HA FGSP + standalone-config-sync, or autoscale, to run commands as super admin via crafted HA packets. It is rated CVSS 3.1 9.0, Known Exploited: No, with fixes in FortiOS 8.0.1, 7.6.8, 7.4.13 and 7.2.14 and a migration to a fixed release for 7.0. It also states: "Virtual Patch named \"FG-VD-61445.0day.\" is available in FMWP db update 26.082" (Fortinet PSIRT FG-IR-26-273, read 19 Nov 2026). The three checks carry over with Fortinet's names: the database on the device must be update 26.082 or later, the patch's action must block and not only log, and the device must see the traffic. Here the traffic is HA packets, and Fortinet says that on FGCP clusters exploitation requires an HA-segment adjacent attack vector, so a sensor that is not on the HA segment sees nothing.
Blind spots and when to revisit
- The sensor path changes: a new route, or a cloud migration that moves the workload past the sensor
- A bypass variant is published
- The date set for the real fix passes
Exploited flaws raise the bar. A finding on CISA's KEV catalog has its own handling, set out in the decision table row for a KEV finding with no patch. Here CISA marks forensic triage as required, so the three addresses Palo Alto ties to its limited exploit attempts (66[.]42[.]72[.]14, 208[.]83[.]239[.]114 and 149[.]248[.]47[.]234) belong in a search of management and firewall logs whatever the signature does.
How long a stand-in may last is on the maximum durations page.
What to record
Control: IPS signature
Finding: CVE-2026-0322 on <firewall or Panorama>, PAN-OS <version>
Filter IDs and action: Threat IDs 510046 and 510047, action <block or reset, not alert>
Content version on the sensor: <version> installed <date> (needs 9159-10287 or later)
Path conditions: management profile on data-plane interface <yes>; certificate replaced <date>; inbound decryption on <yes>; threat prevention on <yes>
Policy export: <file, date>; proof of prevent mode, not a screenshot of the signature list
Block log entry seen: <date or none found on date>
Sensor placement diagram: <link>
Change ticket: <id>
Ends when: fixed build installed or review date <date>
Sources
- Palo Alto Networks advisory CVE-2026-0322 (vendor), read 19 Nov 2026
- OWASP Virtual Patching Cheat Sheet, Virtual Patching Tools, read 19 Nov 2026
- NIST SP 800-53 Rev. 5.2.0, SI-4, read 19 Nov 2026
- NIST SP 800-40 Rev. 4, Guide to Enterprise Patch Management Planning, Sections 2.1 and 2.2, read 19 Nov 2026
- CISA Known Exploited Vulnerabilities catalog, feed version 2026.11.18, read 19 Nov 2026
- Palo Alto Networks, Objects > Security Profiles > Vulnerability Protection (Exceptions tab, Action column), read 19 Nov 2026
- Palo Alto Networks, Device > Dynamic Updates (Currently Installed column), read 19 Nov 2026
- Palo Alto Networks, threat log fields (Type, Subtype, Threat ID, Action values), read 19 Nov 2026
- Fortinet PSIRT FG-IR-26-273 (CVE-2026-126259), read 19 Nov 2026
Last reviewed