SLH-DSA: accept (NULL, 0) message in SignWithRandom/SignDeterministic

[Arpan0995]

Description

Commit 916de9c (#11381) made the sign and verify entry points accept an empty message passed as (NULL, 0). #11055, merged after it, added crypto callback support to wc_SlhDsaKey_SignWithRandom() together with an argument check of its own, and that check rejects msg == NULL for any length. As a result, wc_SlhDsaKey_SignWithRandom() and wc_SlhDsaKey_SignDeterministic(), which calls it, return BAD_FUNC_ARG for (NULL, 0), while wc_SlhDsaKey_Sign() and wc_SlhDsaKey_Verify() accept it. Both commits are in v5.9.4-stable.

This uses the same check as wc_SlhDsaKey_Sign(), so a NULL msg is rejected only when msgSz is not 0. As in Sign() and Verify(), a NULL message is then replaced with a one-byte static stand-in before the crypto callback dispatch, so devices and the hash code never see a NULL pointer. The stand-in is a one-element array to match the Coverity ARRAY_VS_SINGLETON change in #11457. The doxygen for SignDeterministic, SignWithRandom, Sign and Verify, and the source comments for the first two, are updated to match.

Testing

./configure --enable-slhdsa --enable-cryptocb
make
./wolfcrypt/test/testwolfcrypt
./tests/unit.test --api --group slhdsa

and the same with:

./configure --enable-slhdsa --enable-cryptocb --enable-smallstack \
    --disable-shared \
    CFLAGS="-O1 -g -fsanitize=address,undefined -fno-sanitize-recover=undefined" \
    LDFLAGS="-fsanitize=address,undefined"

• wolfcrypt/test/test.c: the SHAKE128F empty-message block in slhdsa_test_param() now also signs with wc_SlhDsaKey_SignDeterministic(NULL, 0) and verifies the result with wc_SlhDsaKey_Verify(NULL, 0). With --enable-cryptocb this also runs under the crypto callback test.
• tests/api/test_slhdsa.c: test_wc_SlhdsaDecisionCoverage() calls SignWithRandom() with (NULL, 0) and a too-small sigSz, which now passes the argument check and returns BAD_LENGTH_E. The existing msg == NULL probe with a nonzero msgSz still expects BAD_FUNC_ARG.

Without the wc_slhdsa.c change, both new checks fail with BAD_FUNC_ARG. With it, testwolfcrypt (including the crypto callback test) and the slhdsa API group pass in both builds. Tested on macOS (arm64) with Apple clang.

Checklist

• added tests
• updated/added doxygen
• updated appropriate READMEs
• Updated manual and documentation

[wolfSSL-Bot]

Can one of the admins verify this patch?

[Copilot]

Copilot review overview

🟢 Approval recommended

The implementation safely aligns API behavior, callback handling, tests, and documentation without unresolved issues.

Review effort: Balanced
Findings: None

What changed in this PR

Enables SLH-DSA randomized and deterministic signing APIs to accept empty messages as (NULL, 0), aligning them with existing sign/verify behavior.

Changes:

• Permits and safely canonicalizes (NULL, 0) messages before callback dispatch.
• Adds round-trip and argument-validation coverage.
• Updates public and source documentation.

File	Description
wolfcrypt/​src/​wc_slhdsa.c	Updates validation and empty-message handling.
wolfcrypt/​test/​test.c	Adds deterministic empty-message round-trip coverage.
tests/​api/​test_slhdsa.c	Tests argument-check behavior for (NULL, 0).
doc/​dox_comments/​header_files/​wc_slhdsa.h	Documents accepted empty-message semantics.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.