Fix benchmark guard macros

[miyazakh]

Fix benchmark guard macros

Summary

• Fix incorrect preprocessor guard for ML-DSA 87 benchmark functions in wh_bench_mod_mldsa.c
• Add missing WOLFHSM_CFG_ENABLE_SERVER guard around wh_Bench_ServerCfgLoop() in wh_bench.c

Changes

File	Description
benchmark/bench_modules/wh_bench_mod_mldsa.c	Replace incorrect guard WOLFSSL_MLDSA_NO_SIGN with WOLFSSL_NO_ML_DSA_87 for ML-DSA 87 benchmark section
benchmark/wh_bench.c	Wrap wh_Bench_ServerCfgLoop() with #if defined(WOLFHSM_CFG_ENABLE_SERVER) guard

Details

Fix typo in ML-DSA 87 guard (wh_bench_mod_mldsa.c)

The wrong macro WOLFSSL_MLDSA_NO_SIGN was used to guard the ML-DSA 87 benchmark section. The correct macro is WOLFSSL_NO_ML_DSA_87, consistent with the guards used for ML-DSA 44 and ML-DSA 65 in the same file.

• Before: #if !defined(WOLFSSL_MLDSA_NO_SIGN)
• After: #if !defined(WOLFSSL_NO_ML_DSA_87)

Without this fix, defining WOLFSSL_NO_ML_DSA_87 to disable ML-DSA 87 support would not exclude the corresponding benchmark functions from compilation.

Guard server benchmark loop (wh_bench.c)

wh_Bench_ServerCfgLoop() uses server APIs (wh_Server_Init, wh_Server_HandleRequestMessage, etc.) and must not be compiled in configurations where server support is disabled. A stub implementation is provided under #else so that callers such as _whBenchServerTask() link cleanly without server support.

• Before: No guard
• After: Full implementation wrapped with #if defined(WOLFHSM_CFG_ENABLE_SERVER); stub (returns WH_ERROR_OK) provided under #else

Impact

• No logic changes; preprocessor guard fixes only.
• Builds with WOLFSSL_NO_ML_DSA_87 will now correctly exclude ML-DSA 87 benchmark functions.
• Builds without WOLFHSM_CFG_ENABLE_SERVER will now link cleanly via the stub implementation.

[Copilot]

Pull request overview

This PR fixes the preprocessor guard that controls compilation of the ML-DSA 87 benchmark functions so it correctly honors the WOLFSSL_NO_ML_DSA_87 feature-disable macro (consistent with the ML-DSA 44/65 benchmark sections and benchmark/wh_bench.c’s registration guards).

Changes:

• Replace #if !defined(WOLFSSL_MLDSA_NO_SIGN) with #if !defined(WOLFSSL_NO_ML_DSA_87) at the start of the ML-DSA 87 benchmark section.
• Align the ML-DSA 87 section’s opening guard with its existing #endif /* !defined(WOLFSSL_NO_ML_DSA_87) */ and with other code that references ML-DSA 87.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

[bigbrett]

I'm confused why in your latest commit you added all of these gates putting identical void casting logic in both cases? We shouldnt add conditional implementations unless it does something.

[miyazakh]

In your earlier comment, you asked, "Do we not also still need WOLFSSL_MLDSA_NO_SIGN here?", so I guarded the sign operation in the same way as wh_Bench_Mod_MlDsa44Sign(), which returns WH_ERROR_NOTIMPL when WOLFSSL_MLDSA_NO_SIGN is defined.

However, since _benchMlDsaSign() does not currently support ML-DSA 65/87, the added guard creates a redundant code path. I applied the same pattern to the other functions for consistency.

Going forward, I see two options:

1. Remove the added macro guards.
2. Extend _benchMlDsaSign(), _benchMlDsaVerify(), and _benchMlDsaKeyGen() to support ML-DSA 65 and 87.

[bigbrett]

based on your original diff, the comment was valid since you removed the dependency on sign, but I see no why you did that. I can merge this now but can you please add a follow up PR introducing benchmark support for all security levels? Thanks!