[hmac,difs] Update DIFs and DIF-based TLTs

[gdessouky]

Description

At the moment, HMAC DIFs and drivers configure HMAC by default in SHA-2 256 and 256-bit key. To support TLTs configuring different digest sizes and key lengths, the DIFs need to be updated and the corresponding documentation. Additionally, we need to exercise reading the HMAC error code for invalid configuration.

[ballifatih]

My plan is to focus on extending DIFs and adding a top-level test for streaming. Just mentioning this here, so we do not duplicate the effort.

[ballifatih]

At the moment, we are using TLTs implemented on top of cryptolib HMAC driver to get nightly results. It would make sense to implement these tests on top of DIFs instead of the driver, but I think that is not M4 priority.

We have two relevant top-level tests, one is testing oneshot calls and the other one is testing multistream (i.e. streams that are running in parallel and interleaving each other with context switching).

Relevant PRs:

• [HMAC, cryptolib] Connect all SHA-256/384/512 and HMAC-256/384/512 to HMAC driver. #23335
• [top-level] Update HMAC top level test plan. #23471
• [earlgrey, hmac] Add TLT for SHA-2/HMAC multistream #23494

I would update the milestone as M5 or later.

[gdessouky]

Labeling this as M5 since at M4 we exercise the new features at block-level and test one-shot and streaming HMAC with KAT test vectors at the top-level using the driver-based TLTs referenced above by @ballifatih.

cc @martin-velay

[vogelpi]

In my view, if we have TLTs using these hardware features through the cryptolib, we don't need to write further TLTs for these features. Ideally, the DIF can still be extended provide those interfaces.

However, both tasks don't sound like a M5 item to me. What we care about is the coverage of these features at the top level (and block level) and my understanding is that we have that coverage now already.

[vogelpi]

So in my view we can de-prioritze this.

[vogelpi]

From an offline discussion between @gdessouky and @martin-velay : "this will enable us to try out at top-level different scenarios with context switching" - 4 days estimated effort.

[vogelpi]

Discussed in triage meeting to leave this in.

[martin-velay]

Moved to M7 backlog as this is not relevant for the V3 signoff