glm-5.3-flash
9ed3e8c83d
- Cargo.toml: alkcall 0.8.1 -> 0.9.0 (crates.io, the stub-sweep release) and alktunnels 0.2.1 -> 0.3.0 (its dependency-major ride); 0.1.0 -> 0.2.0 (semver-incompatible dep bumps + public alkcall re-exports -> minor, matching the alktunnels 0.2.1 -> 0.3.0 ride) - fuzz/shared/Cargo.toml: alkcall 0.9.0 (one alkcall version across the path graph) - No code change: zero contact with the 0.9.0 breaking surface (no marked-pub registration, no control-handle subscription, no resource enumerator use, no generic op/register announcements; Establishment::default() is the only Establishment contact, in the session gate) - CHANGELOG: 0.2.0 section + compare links - README: the 0.2.0 note - docs/architecture/overview.md: dep tree + feature table alkcall 0.9.0 / alktunnels 0.3.0 - AGENTS.md: alkcall refs — v0.8.x/0.8.0 pin -> v0.9.x/0.9.0 Verification: cargo test (145 default / 171 all-features), fuzz/shared corpus replay (3/3), clippy --all-targets -D warnings (native + wasm32-unknown-unknown), fmt --check, wasm32 check, cargo doc, and cargo publish --dry-run --allow-dirty all green
alksocks fuzzing
libFuzzer targets for alksocks' wire surfaces (see
docs/research/fuzzing.md for the full rationale and campaign plan).
Layout
fuzz_targets/— one binary per target; thinfuzz_target!wrappers.shared/— the invariant logic, as a plain library so normalcargo test(stable toolchain) can replay the committed corpora through the same invariants (fuzz/shared/src/*.rs#[cfg(test)]modules; quinn's CI pattern). The fuzz binaries are nightly-only; the shared crate is stable-clean.corpus/<target>/— committed hand-made seeds (regenerate withpython3 fuzz/gen_fuzz_seeds.pyfrom the repo root). Grown corpora and artifacts are gitignored.run-detached.sh— mandatory runner for agent sessions: wrapscargo fuzz runinsetsid+nohup+ log redirection so an OOM in a target can never take down the agent host (research doc §5).json.dict— JSON token dictionary for the params target.
Targets
| Target | Drives | Invariants |
|---|---|---|
datagram_codec |
DatagramReader::feed / frame / frame_inner (src/wire.rs) |
no-panic; infallible reassembly with the decode-error partition; empty-payload F-2 layering; bad-ATYP loudness; splitting transparency; round-trips; the joint bound (incl. the 65766 shape); the decoder-state retained-bytes bound |
rfc1928_codec |
the forked RFC 1928 core (src/rfc1928/): read_address, the server typestate handshake, the RFC 1929 sub-handshake, the RFC UDP header codec, ReplyError::from_u8 |
no-panic on any byte sequence, both protocol directions; typed failures only (the F-1 class); round-trips; the full-domain reply table |
params_parse |
Socks5Params / socks5_open_spec() / the reason surface (src/params.rs) |
no-panic; the trivial-params pin; the spec is a constant (the wire-stable ADR-001 surface + the schema/typed-view drift guard); the pass-through reason-extraction posture |
Commands
cargo fuzz build and cargo fuzz run must be executed with fuzz/
(or deeper) as the working directory so rustup selects the pinned
nightly toolchain — the detached runner handles that itself.
# build (nightly, pinned by rust-toolchain.toml inside fuzz/; run from fuzz/)
cargo fuzz build
# agent sessions: detached campaign (never foreground; CWD-independent)
FUZZ_RUNTIME_SECS=600 fuzz/run-detached.sh datagram_codec -max_len=65536
FUZZ_RUNTIME_SECS=600 fuzz/run-detached.sh rfc1928_codec -max_len=4096 -use_value_profile=1
FUZZ_RUNTIME_SECS=600 fuzz/run-detached.sh params_parse -dict=json.dict -use_value_profile=1
# corpus replay through the invariants (stable toolchain, no nightly)
cargo test --manifest-path fuzz/shared/Cargo.toml
# coverage
cargo fuzz coverage <target>
The fuzz workspace is excluded from the main workspace and from the
published package (exclude in the root Cargo.toml); it pins its own
nightly toolchain via rust-toolchain.toml and does not affect the
crate's stable MSRV.