CVE-2022-36765 is a high severity (CVSS 7.0) memory safety flaw in CreateHob() in TianoCore EDK2, the open source UEFI firmware base. EDK2 releases up to 202311 are affected and the advisory lists no patched release. Apply the vendor patch, or firmware updates from your hardware supplier.
What's affected
The TianoCore GitHub advisory lists the affected component as UefiPayloadPkg in EDK2, affected versions up to and including 202311, and patched versions as None. It states that patch files are available through the TianoCore Bugzilla entry https://bugzilla.tianocore.org/show_bug.cgi?id=4166 and that the fix was planned for the February 2024 EDK2 release. I did not confirm that the release shipped, so check the current EDK2 tree and release notes.
The record scores the flaw 7.0 (High) with vector CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:C/C:L/I:L/A:H: local access, high complexity, low privileges, and a changed scope. The weakness classes are CWE-680 (Integer Overflow to Buffer Overflow) and CWE-119 (Improper Restriction of Operations within the Bounds of a Memory Buffer).
EDK2 is a code base, not a product you install. It sits inside the firmware of many servers, laptops and virtual machine platforms, so the practical question is which firmware builds you run and whether your hardware vendor has merged the fix.
How to check if you're exposed
You will not find EDK2 in a package manager on most machines, so the check is about firmware and virtualization.
- List the hardware and virtual machine platforms in your fleet that boot through UEFI and use TPM measured boot.
- Ask each hardware vendor, or read its security bulletins, for the firmware version that includes the TianoCore fix for Bugzilla 4166.
- If you build your own firmware from EDK2, check whether your source tree includes the fix for CreateHob().
On a Linux host you can read the firmware version the machine reports, then compare it with your vendor's bulletin:
sudo dmidecode -s bios-version
sudo dmidecode -s bios-release-date
If you run virtual machines with OVMF, the EDK2 based firmware for QEMU and KVM, ask your distribution whether it has backported the fix. Distribution announcements for this CVE exist in the record's references, though I did not read them.
What to do now
- Inventory firmware and virtual machine firmware that is built on EDK2.
- Check vendor bulletins for a firmware update that includes the TianoCore fix referenced by Bugzilla 4166.
- If you maintain your own EDK2 build, apply the patch from the Bugzilla entry or move to an EDK2 release that includes it.
- Limit who can write to the EFI system partition and who can attach untrusted disks or images to machines that matter, since the input comes from disk.
- Re-check after your vendors publish firmware updates, and record the build you end up on.
Why it matters
CreateHob() builds a HOB, a Hand Off Block that passes data between early boot phases. It rounds the requested length up to a multiple of eight and stores the result in a 16 bit value, with no check that the rounding did not overflow. A request near the top of the range can come back smaller than asked, and code that then reads or writes the HOB may run past its end.
The advisory is candid about exploitability. It calls it tricky, because the attacker would have to trigger the condition in the PEI phase, but notes that the function is called often. That balance explains the score: a serious class of bug with a narrow route. The record's KEV flag is false and the advisory does not report exploitation.
For defenders the lesson is the same as for the other two EDK2 entries in this series: firmware code is code, it has integer overflow bugs, and your update path for it needs an owner.
Quick check
Why is this scored the same as the others if it is harder to reach?
The CVSS vector in the record is the same for all three, local and high complexity. The advisory text, not the score, tells you this one is harder to trigger.
Which EDK2 package is involved?
The advisory names UefiPayloadPkg.
Sources
Safeguard can help you see this across many sites or repositories at once. Paste a manifest into the free in-browser dependency scan at gold.safeguard.sh/scan or look the issue up in the Gold directory to see how the record is tracked.
Last reviewed: 19 October 2026.