component_test_full_no_bignum doesn't actually disable bignum

[gilles-peskine-arm]

As of #9172, component_test_full_no_bignum actually enables all crypto because it has MBEDTLS_PSA_CRYPTO_CONFIG enabled. For example, PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY is enabled by default, so MBEDTLS_RSA_C and MBEDTLS_BIGNUM_C are auto-enabled to implement that.

The goal of this task is:

• Fix component_test_full_no_bignum to do what it actually claims. For 3.6, it's enough to disable MBEDTLS_PSA_CRYPTO_CONFIG. For development leading to 4.0, we need to disable all asymmetric crypto through both MBEDTLS_xxx and PSA_WANT_xxx.
• In component_test_full_no_bignum, check that mbedtls_bignum_xxx functions aren't getting included in the build.
• Review other similar components.

Alternative goal: remove component_test_full_no_bignum because it's redundant with test-ref-configs on configs/config-symmetric-only.h?

[ronald-cron-arm]

I've fixed that in #9185.

[mpg]

Alternative goal: remove component_test_full_no_bignum because it's redundant with test-ref-configs on configs/config-symmetric-only.h?

It is especially redundant with component_test_psa_crypto_config_accel_ecc_no_bignum(). Specifically, component_test_full_no_bignum was added as part of preparation work for that component, so now that we have component_test_psa_crypto_config_accel_ecc_no_bignum() I don't think component_test_full_no_bignum really serves a purpose any more.

[ronald-cron-arm]

Alternative goal: remove component_test_full_no_bignum because it's redundant with test-ref-configs on configs/config-symmetric-only.h?

It is especially redundant with component_test_psa_crypto_config_accel_ecc_no_bignum(). Specifically, component_test_full_no_bignum was added as part of preparation work for that component, so now that we have component_test_psa_crypto_config_accel_ecc_no_bignum() I don't think component_test_full_no_bignum really serves a purpose any more.

Thanks Manuel. I am removing it in #9185.

[mpg]

Looks like this was fixed by #9185 - I'm not seeing full_no_bignum in development any more. So I'm closing this issue. Please shout if I missed something.

[gilles-peskine-arm]

I'm reopening because this is moot in development, but still a bug in 3.6.