[yosuke-wolfssl]

Problem

wolfSSL_TXT_DB_create_index() requires a non-NULL cmp callback but discards it, and wolfSSL_TXT_DB_get_by_index() delegates to wolfSSL_lh_retrieve(), which treats any hash match as a hit.

Fix (src/conf.c)

• WOLFSSL_TXT_DB gains cmp_fn[WOLFSSL_TXT_DB_MAX_FIELDS], appended after hash_fn.
• wolfSSL_TXT_DB_create_index() stores cmp in cmp_fn[field].
• wolfSSL_TXT_DB_get_by_index() walks the rows itself and returns a row only when its cached hash matches and cmp_fn(row, value) == 0. It also rejects a NULL value and an index that has no compare callback.

Closes f-8864.

Tests

test_wolfSSL_TXT_DB() re-creates index 3 with a constant hash, so every row collides, and checks:

• a lookup of 12DA returns the 12DA row (before the fix it returned the first row, 12F8);
• a lookup of FFFF returns NULL.

Verification

• The new check fails before the fix and passes after it.
• ./tests/unit.test (--enable-opensslall --enable-opensslextra): 0 failed, 1461 passed.
• ASan + UBSan clean on test_wolfSSL_TXT_DB.
• GCC on Ubuntu 24.04 with -Werror: build clean, test passes.

[Copilot]

🟢 Approval recommended

No active review issues block approval of the TXT_DB hash-collision fix.

No open issues 🎉

What changed in this PR

This PR fixes hash-collision handling in wolfSSL’s OpenSSL-compatible TXT_DB lookups by requiring both a matching hash and a successful comparison callback.

Changes:

• Stores a comparison callback for each TXT_DB index.
• Checks row equality during lookup and rejects NULL values or missing callbacks.
• Adds collision regression tests for matching and absent values.

File	Description
wolfssl/​openssl/​txt_db.h	Adds per-index comparison callback storage.
tests/​api.c	Tests successful and unsuccessful lookups with colliding hashes.
src/​conf.c	Stores comparison callbacks and uses them to verify lookup matches.

_{Review effort: Balanced}

[github-actions]

MemBrowse Memory Report · Explore what changed →

gcc-arm-cortex-m4-openssl-compat

• FLASH: .text +128 B (+0.0%, 800,804 B / 1,048,576 B, total: 76% used)

zephyr-frdm_rw612

• FLASH: log_const_area -8 B, rodata -8 B, text +1,592 B (+0.6%, 284,040 B / 67,108,864 B, total: 0% used)

No memory changes detected for:

• gcc-arm-cortex-m0plus
• gcc-arm-cortex-m3
• 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-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
• stm32-sim-stm32h753

---

_{Catch memory regressions in CI with MemBrowse. Free for open source.}

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #12034

Scan targets checked: wolfssl-src, wolfssl-bugs
Coverage: 3 of 3 in-scope changed file(s) opened by the reviewer

Fenrir result: Approved ✅

No new issues found in the changed files.

Advisory only — this automated result does not count as a GitHub approval.

Review tier: Max

[yosuke-wolfssl]

@claude review

[claude]

I reviewed this PR and didn't find any bugs. It adds a field to the public WOLFSSL_TXT_DB struct in wolfssl/openssl/txt_db.h, which changes the struct layout for OPENSSL_ALL builds, so a maintainer should still confirm that the change is acceptable.

What was reviewed:

• The new lookup loop in wolfSSL_TXT_DB_get_by_index. When the hash function changes, the existing sync loop resets data->hash to 0, so cached hashes don't go stale when an index is re-created. The new test relies on this when it switches from TXT_DB_hash to TXT_DB_hash_const.
• wolfssl_sk_node_get_data is static in src/ssl_sk.c. It is still visible here because ssl.c includes ssl_sk.c before conf.c. wolfSSL_TXT_DB_read zeroes the struct with XMEMSET, so cmp_fn starts out NULL.
• Callers that set hash_fn[idx] directly without calling create_index will now get NULL from every lookup. That only affects code that bypasses the API, so I don't consider it a bug.

Extended reasoning...

The change touches three files. In src/conf.c, the TXT_DB index creation now stores the compare callback, and the lookup now requires both a hash match and a compare match. wolfssl/openssl/txt_db.h gains a public struct field, and tests/api.c adds a test where every hash collides. It doesn't touch crypto, auth or network-parsing paths. The logic is small and follows the existing wolfSSL_lh_retrieve walk, and the test covers it. I deferred for two reasons: the public struct layout (ABI) change needs a maintainer's call, and the earlier reviews from other reviewers have no later commit, so I couldn't confirm they were addressed.