PQC efforts integration for OpenEmbedded

[embetrix]

Hi,

I’ve noticed we are both working on Post-Quantum Cryptography integration to the Yocto ecosystem !

I recently completed a significant rework of meta-oqs (formerly meta-pqc) to align with the latest releases.

I also implemented a more modular/seamless integration on OpenSSL and created several demos.

I noticed some overlapping patches and updates between our layers recently.
Given that both meta-oqs/meta-quantum-safe are already on the Layer Index, I think the community would really benefit from us coordinating our efforts rather than having fragmented PQC layers.

what do you think ?

Best regards

[ajlennon]

Good to hear from you. Totally agree - I was informed you'd set up meta-oqs. Whilst I started with liboqs support I want meta-quantum-safe to be wider support for "all things QS for embedded". You'll see I recently updated this layer to build the latest liboqs with CI build and test and so forth. How do you think we can collaborate?

fwiw - I also really think we want to be looking at Zephyr support somehow...

[embetrix]

Thanks for the feedback.

To be honest I see OQS primarily as a bridge. While it’s great for experimentation from another side it isn’t quite production-ready and will never be :-( especially since OpenSSL 3.5+ has already started integrating PQC algorithms natively.

I see the real value of such layer as supporting developers who are currently stuck on Yocto LTS releases and cannot yet move to the latest versions but still need to evaluate or implement PQC today.

Beyond the existing demos for Nginx, MQTT and OpenVPN, I’m looking into expanding the scope toward smaller MCUs (like STM32) and exploring TrustZone integration to see how these algorithms behave in more constrained, secure environments.

I think Zephyr would also greatly benefit from that

[ajlennon]

To be honest I see OQS primarily as a bridge. While it’s great for experimentation from another side it isn’t quite production-ready and will never be :-( especially since OpenSSL 3.5+ has already started integrating PQC algorithms natively.

Yes that was more or less what I was hearing at FOSDEM. And that OpenSSL 3.5+ is now integrating PQC

I have various questions I'd like to look at, e.g.

• accelerated silicon and lack of availability. What are we going to be seeing in TPMs / Secure Elements and when?
• what are the memory and performance requirements of QS support (I see multiple target support in the CI being useful here, perhaps with some reporting tool akin to benchmark testing tools)
• what are we going to find in terms of weaknesses in these algorithms as they are tested in the field.

I think this space can be productively used for those kinds of purposes

[embetrix]

Silicon manufacturers are starting to integrate PQC but since the final standards only dropped in 2024, it’s still early days for TPMs and PKCS#11. We’re in a transition period.

Regarding your points:

I've optimized liboqs for AARCH64 and x86-64 and the results for TLS handshake were quite surprising :
https://github.com/orgs/open-quantum-safe/discussions/2372

Weaknesses: The consensus is definitely to adopt hybrid key exchange at early stages to hedge against any future breaks

[ajlennon]

Silicon manufacturers are starting to integrate PQC but since the final standards only dropped in 2024, it’s still early days for TPMs and PKCS#11. We’re in a transition period.

Yes, we are chatting to NXP as we have been using their SE05x parts for some years and it looks like the new part in the range is targetted at what they seem to be calling "quantum agility" or the ability to update in the field as nobody really seems to know exactly what is coming.

I am hearing whispers that there is a problem with them not actually being able to fit the code into the part so I am interested to see where this is going...

[embetrix]

I have evaluated the SE05x chip (not really a TPM but a secure storage) and its software stack some years ago, it's already very weak and not recommended for any serious usage...