Fix odd-length CertificateRequest signature_algorithms acceptance

[yosuke-wolfssl]

Description

The TLS 1.2 CertificateRequest parser bounds-checked the supported_signature_algorithms vector but never verified that its length is an even multiple of the 2-byte SignatureAndHashAlgorithm element size. PickHashSigAlgo iterates while (i+1) < hashSigAlgoSz, so an odd trailing byte was silently ignored — a malformed vector (e.g. length 3: one valid pair + 1 junk byte) could select an algorithm and return success instead of being rejected.

This adds the missing parity check, matching what the TLS 1.3 signature_algorithms extension parser already does (TLSX_SignatureAlgorithms_Parse in src/tls.c).

Addressed by f_4890.

Changes

• src/internal.c — DoCertificateRequest: reject TLS 1.2 supported_signature_algorithms lengths not divisible by HELLO_EXT_SIGALGO_SZ with BUFFER_ERROR, before PickHashSigAlgo. The check sits ahead of the client-cert/key handling so the malformed message is rejected consistently even when no local client certificate is configured.
• tests/api/test_tls.c / tests/api/test_tls.h — new regression test test_tls12_certreq_odd_sigalgs. It drives a real TLS 1.2 memio handshake (server requesting a client cert), locates the server's CertificateRequest, shrinks the sig-algs vector by one byte while decrementing the record/handshake/sig-algs length fields so the message stays self-consistent (only the length parity is wrong), and asserts the client rejects with BUFFER_ERROR.

Testing

• New test fails without the src/internal.c fix (client silently accepts the odd-length vector and continues the handshake) and passes with it — confirming it catches the regression rather than passing incidentally.
• Full tests/unit.test suite passes: Success for all configured tests (1428 passed, 0 failed).

[Copilot]

Pull request overview

This pull request hardens the TLS 1.2 CertificateRequest parser by rejecting malformed supported_signature_algorithms vectors whose length is not an even multiple of the 2‑byte SignatureAndHashAlgorithm element size, aligning behavior with existing TLS 1.3 parsing.

Changes:

• Reject odd-length TLS 1.2 supported_signature_algorithms in DoCertificateRequest() with BUFFER_ERROR before algorithm selection.
• Add a regression test that performs a real TLS 1.2 memio handshake, corrupts the CertificateRequest sig-algs vector length parity while keeping length fields self-consistent, and asserts the client fails with BUFFER_ERROR.
• Register the new test in the TLS API test list.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
src/internal.c	Adds an explicit modulo check to reject malformed TLS 1.2 sig-algs vectors before calling PickHashSigAlgo().
tests/api/test_tls.c	Adds a memio-based regression test that mutates a server CertificateRequest to have an odd sig-algs vector length and expects BUFFER_ERROR.
tests/api/test_tls.h	Declares and registers the new regression test in the TLS test group.

---

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

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #10630

Scan targets checked: none
Failed targets: wolfcrypt-rs-bugs, wolfssl-bugs, wolfssl-src

⚠️ Review incomplete — one or more scan targets failed before findings could be produced. See the Fenrir PR review detail page for logs.

[github-actions]

MemBrowse Memory Report

gcc-arm-cortex-m3

• FLASH: .text +4 B (+0.0%, 120,909 B / 262,144 B, total: 46% used)
  No memory changes detected for:
• gcc-arm-cortex-m0plus
• gcc-arm-cortex-m4
• gcc-arm-cortex-m4-baremetal
• gcc-arm-cortex-m4-crypto-only
• gcc-arm-cortex-m4-dtls13
• gcc-arm-cortex-m4-min-ecc
• gcc-arm-cortex-m4-openssl-compat
• gcc-arm-cortex-m4-pkcs7
• gcc-arm-cortex-m4-pq
• gcc-arm-cortex-m4-rsa-only
• gcc-arm-cortex-m4-sp-math
• gcc-arm-cortex-m4-tls12
• gcc-arm-cortex-m4-tls13
• gcc-arm-cortex-m7
• gcc-arm-cortex-m7-pq
• gcc-arm-cortex-m7-tls13
• linuxkm-pie
• linuxkm-standard
• stm32-sim-stm32h753

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #10630

Scan targets checked: none
Failed targets: wolfcrypt-rs-bugs, wolfssl-bugs, wolfssl-src

⚠️ Review incomplete — one or more scan targets failed before findings could be produced. See the Fenrir PR review detail page for logs.