Manuel Pégourié-Gonnard

@tom-cosgrove-arm can you have a look at the rebase and re-approve if you're happy with it?

> I created script for generating test vectors for FFDH key agreement , added tests for psa_raw_key_agreement and everything works except tests for 1024 bytes keys. In this case I'm...

> We don't yet have accelerator support for key agreement. Are you sure about this? Either we do, or [this component](https://github.com/Mbed-TLS/mbedtls/blob/development/tests/scripts/all.sh#L2232) is not doing at all what I think it's...

Generally speaking, if a test that's unrelated to the PR passed on one CI and fails on the other, we can ignore it. (If the test is possibly related to...

> Ah, no, you're right, we do now. But indeed, this PR is already big enough, let's not expand the scope to driver support. Should we create a follow-up issue...

See #7568 and #7569.

This can be rebased now that #6010 has been merged, and also could address https://github.com/Mbed-TLS/mbedtls/pull/6010#pullrequestreview-1420180777

> * `ECDH` related client/server code is guarded using `MBEDTLS_ECDH_C`. What should be used in case of `FFDH` `MBEDTLS_DHM_C` (no functions from this module are used here) ? As a...

> > * `ECDH` related client/server code is guarded using `MBEDTLS_ECDH_C`. What should be used in case of `FFDH` `MBEDTLS_DHM_C` (no functions from this module are used here) ? >...

Note: if we go with renaming existing fields & functions, and making some buffers dynamic, I think we should consider making that part a separate PR. I find it a...