Bluetooth: Host: Add encrypted advertising

[theob-pro]

Create a new Bluetooth library and add Encrypted Advertising Data to it.

Encrypted Advertising Data is a new feature from the Bluetooth Core Specification 5.4. It provides a way to communicate encrypted data in advertising, scan response and EIR packets. To do that it introduce a new advertising data type called Encrypted Advertising Data. Also, it introduce a new characteristic called Encrypted Data Key Material, this provides a way to share the key material.

The library add two main functions bt_ead_encrypt and bt_ead_decrypt.

Also, add a new BabbleSim test that use the sample data from the Supplement to the Bluetooth Core Specification to validate the encryption, decryption and authentication.

Related issue: #55081

[Thalley]

The second commit should be tests: bluetooth:

Why is this being added as a library, instead of just an API in bluetooth.h?

[jori-nordic]

Why is this being added as a library, instead of just an API in bluetooth.h?

The host is pretty bloated and tries to be too nice to the user. This pushes functionality that shouldn't be in the host to be inside the host (settings usage, a bunch of gatt functionality, adv resume comes to mind). That functionality then makes use of the internal state and functions, because why not, it's more efficient and that results in:

• more bugs, because more complex and internal state is used all over the place
• the API being not that nice to use in practice, because the user still has to be pretty familiar with the host's inner workings to use the 'user-friendly' functionality

That's why we'd like to slim down the host to a maintainable core + some libraries on top.
Encrypted advertising can be implemented fully outside of the host, that's why we're making it a library.

[hermabe]

The current tests only check that the advdata encrypts and decrypts back to itself. It would be nice to have some tests that checks that the encrypted data is correct as well. I suggest we use the "Encrypted Advertising Data Sample Data" from the CSS.

[jori-nordic]

but we should also look at what we already have and see if it makes sense to group this new functionality with existing Bluetooth sub-modules that can also be considered libraries (in the sense that there'd be multiple "library submodules" under the same directory.

This sounds like it should be done in a separate PR. Having this lib/ folder will make it easier to move functionality out of the 'core' host in the near future anyway.

[jhedberg]

@theob-pro I think it might make sense to rename the header file simply to ead.h, so that it's in line with the API naming. Probably same with the c-file. Application developers will be primarily referencing the generated documentation, so I don't think this will make the API any less discoverable for them.

[theob-pro]

Rebased to resolve file conflict (tests/bluetooth/bsim/host/compile.sh).

[theob-pro]

Rebased to resolve file conflict (tests/bluetooth/bsim/host/compile.sh).

[Thalley]

My remaining comments are about undocumented/required formatting changes, and one comment about a LOG_WRN that I think should be LOG_DBG, but it's not bad enough that I will block this PR anymore :)

[jori-nordic]

I agree, let's get this in, and @theob-pro and @Thalley can discuss the small stuff in a new PR.

[jhedberg]

Sorry for being slow to review this. I've tried to keep my comments restricted to things which require more effort/pain to fix later in the game, i.e. the API details.