retoorandClaude Opus 4.8
79ae6f27da
Response to incident WA-2026-10-27 (the std::vector max_size() crash). Deep research archived under research/safe-cpp-practices (8 Tier 1, 37 Tier 2, 2 conclusions) on production-safe C++ that keeps development speed: OpenSSF compiler hardening, the C++ Core Guidelines bounds profile, integer-overflow practice, libFuzzer/sanitizer CI, the sanitizer overhead ladder, clang-tidy ratchets. The research prescribed an ASan+UBSan+_GLIBCXX_ASSERTIONS build of the parsers. Standing it up immediately caught a second, independent undefined-behavior bug that two manual reviews had missed: a protobuf varint loop that shifted by up to 70 bits (shift >= 64 is UB) on malformed input. Fixed and guarded by a deterministic test and the sanitizer net. - src/whatsapp/wa_proto.cpp: guard the wire-0 varint shift (the UB). - tests/whatsapp_test.cpp: add an overlong-varint hostile case. - tools/san-whatsapp.sh: the standing sanitizer net for detz::wa parsers. - WaClient frame capture: a frame that fails to decode/dispatch has its exact bytes recorded to <home>/whatsapp/last-error.log (bounded), so the still unreproduced live <iq> failure becomes a one-paste deterministic fix. Wired from the wizard and the whatsapp_send tool. - docs/cpp-safety.md: the project-wide playbook and design decisions (four layers: overflow-safe checked-reader parsing, cheap release hardening, the sanitizer test net, capture). docs/whatsapp.md section 13.6 records the follow-up and the honest open item. AGENTS.md makes it a standing rule; cap.md and mcp.md inherit it for their wire parsers. Verified: build, whatsapp_test (incl. hostile+fuzz), make lint, and tools/san-whatsapp.sh all green; research audit PASS. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
51 lines
2.1 KiB
Markdown
51 lines
2.1 KiB
Markdown
# Coverage-guided fuzzing of parsers with libFuzzer + ASan/UBSan in CI
|
|
|
|
## Metadata
|
|
|
|
| Field | Value |
|
|
|---|---|
|
|
| Institute | Molodetz Institute — Research & Retrieval Division |
|
|
| Subject | safe-cpp-practices |
|
|
| Document type | Source Dossier |
|
|
| Tier | 1 |
|
|
| Source title | Coverage-guided fuzzing of parsers with libFuzzer + ASan/UBSan in CI |
|
|
| Source URL | https://appsec.guide/docs/fuzzing/c-cpp/libfuzzer/ |
|
|
| Source publisher / author | Trail of Bits (appsec.guide); debugg.ai |
|
|
| Source type | vendor docs |
|
|
| Source date (published) | Unknown — not stated |
|
|
| Date researched | 2026-10-27 |
|
|
| Researched by | tai (agent), model: Claude Opus 4.8, on behalf of retoor |
|
|
| Research method | web search synthesis |
|
|
| Confidence | High |
|
|
| Status | Active |
|
|
|
|
## Summary
|
|
|
|
The standard way to harden an untrusted binary parser: a fuzz harness per entry point, built with -fsanitize=fuzzer,address,undefined and -fno-sanitize-recover, run short on every push and long nightly, with a regression corpus of past crashes rerun on every build.
|
|
|
|
## Key findings
|
|
|
|
- Build with `-fsanitize=fuzzer,address,undefined` and `-fno-sanitize-recover` so a finding aborts rather than warns.
|
|
- One harness per untrusted-input entry point, with bounded input sizes.
|
|
- Keep a regression corpus of past crashing inputs and rerun it on every build.
|
|
- 'No crashes' only means something if the harness reaches the risky code, so track coverage.
|
|
- ASan adds ~2-4x runtime; it is a test tool, not a release one.
|
|
|
|
## Pros / strengths
|
|
|
|
Concrete, matches industry practice; the regression-corpus rule is exactly what turns an unreproducible bug into a permanent test.
|
|
|
|
## Cons / limitations / caveats
|
|
|
|
libFuzzer is now maintenance-only at LLVM (successor: Centipede); the CI pattern is unaffected but tool choice may shift.
|
|
|
|
## Relevance to tai
|
|
|
|
detz adopts the pattern at its scale: a deterministic hostile-input test plus a random-input fuzz in whatsapp_test, run under ASan+UBSan via tools/san-whatsapp.sh, and a captured-frame file as the regression corpus seed.
|
|
|
|
## Changelog
|
|
|
|
| Date | Change |
|
|
|---|---|
|
|
| 2026-10-27 | Initial dossier created (web search synthesis). |
|