Fix Dilithium signing when WC_DILITHIUM_CACHE_MATRIX_A is enabled

[embhorn]

Description

dilithium_sign_with_seed_mu() allocated matrix A into a local variable a, then immediately overwrote it with a = key->a

Fixes #10383

Testing

Added dilithium_sign_cache_alloc_test

WC_DILITHIUM_CACHE_MATRIX_A is enabled in Debian RPM tests

Checklist

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

[Copilot]

Pull request overview

This PR fixes the Dilithium/ML-DSA signing path when matrix-A caching is enabled, addressing the cached-matrix allocation bug described in #10383. It updates the core signing implementation and adds a regression test in the wolfCrypt test suite to keep this cache-allocation path covered.

Changes:

• Fix dilithium_sign_with_seed_mu() so the cached matrix-A allocation is stored on key->a instead of a transient local pointer.
• Zero-initialize the newly allocated cache buffer before use, matching other cached-allocation paths.
• Add dilithium_sign_cache_alloc_test() and run it for ML-DSA 44/65/87 to exercise the sign-after-cache-reset path.

Reviewed changes

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

File	Description
wolfcrypt/src/dilithium.c	Fixes cached matrix-A allocation in the Dilithium signing path.
wolfcrypt/test/test.c	Adds a regression test that clears the cache, signs, and verifies for all supported ML-DSA levels.

---

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

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #10400

Scan targets checked: wolfcrypt-bugs, wolfcrypt-src

No new issues found in the changed files. ✅

[github-actions]

MemBrowse Memory Report

gcc-arm-cortex-m4

• FLASH: .rodata.wolfSSL_ERR_reason_error_string.str1.1 +35 B, .text +448 B (+0.3%, 197,365 B / 262,144 B, total: 75% used)

gcc-arm-cortex-m4-baremetal

• FLASH: .text +192 B (+0.3%, 64,505 B / 262,144 B, total: 25% used)

gcc-arm-cortex-m4-min-ecc

• FLASH: .text +192 B (+0.3%, 60,091 B / 262,144 B, total: 23% used)

gcc-arm-cortex-m4-tls12

• FLASH: .text +448 B (+0.4%, 120,785 B / 262,144 B, total: 46% used)

[Frauschi]

As ML-DSA is now added to Wconversion testing, you will need type casts on the XMALLOC() and XMEMSET() calls for the params->aSz argument.