Pass Authenticators

This repository contains several authenticators suitable for FRAME-Contrib's pallet-pass.

Workspace

This workspace contains the following crates:

• pass-authenticators-webauthn: This authenticator uses WebAuthn Credentials as devices, and validates its assertions.
• pass-authenticators-webauthn-verifier: A no_std verifier for WebAuthn ES256 (P-256) signatures, used by pass-authenticators-webauthn.
• pass-authenticators-substrate-keys: This authenticator uses Substrate-compatible public keys as devices, and validates its signatures.
• pass-authenticators-ethereum: This authenticator uses Ethereum accounts as devices, and validates their personal_sign (EIP-191) signatures.
• pass-authenticators-bitcoin: This authenticator uses Bitcoin keys as devices, and validates their signed messages (BIP-137).
• pass-authenticators-nostr: This authenticator uses Nostr keys as devices, and validates their BIP-340 Schnorr signatures.
• pass-authenticators-solana: This authenticator uses Solana wallets as devices, and validates their Ed25519 signatures.
• pass-authenticators-ssh: This authenticator uses SSH Ed25519 keys as devices, and validates their SSHSIG signatures.

Every authenticator has a runtime feature (on by default) with the pallet-pass integration. Without it, a crate only exposes its credential types, with the same SCALE encoding, so clients can build them without the Substrate runtime stack.

Versions

The crates are published to crates.io and released in lockstep: they share one version, one CHANGELOG.md and one vX.Y.Z tag. Each major line of this repository pairs with one major line of frame-contrib:

pass-authenticators	frame-contrib	polkadot-sdk	Branch
1.x	2.x	stable2606	release/frame-contrib-v2
2.x	3.x	stable2606	main

Depend on the same version of every pass-authenticators-* crate, and on the matching fc-* version:

[dependencies]
pass-webauthn = { package = "pass-authenticators-webauthn", version = "2.0.0-pre.1", default-features = false, features = ["runtime"] }
pass-substrate-keys = { package = "pass-authenticators-substrate-keys", version = "2.0.0-pre.1", default-features = false, features = ["runtime"] }

See RELEASING.md for how releases are made, and CONTRIBUTING.md for PR titles and what counts as a breaking change.

Verification weights and benchmarks

fc-pallet-pass charges what verifying an attestation (on register/add_device) or a credential (on every extrinsic authenticated with PassAuthenticate) costs through verification_weight, which each authenticator implements from its own benchmarks:

Authenticator	Attestation	Credential
pass-authenticators-webauthn	verify_attestation(c, a)	verify_credential(c, a) (includes the P-256 signature check)
pass-authenticators-substrate-keys	verify_attestation_{sr25519,ed25519,ecdsa,eth}()	verify_credential_{sr25519,ed25519,ecdsa,eth}()
pass-authenticators-ethereum	verify_attestation()	verify_credential()
pass-authenticators-bitcoin	verify_attestation_{compressed,uncompressed}()	verify_credential_{compressed,uncompressed}()
pass-authenticators-nostr	verify_attestation()	verify_credential()
pass-authenticators-solana	verify_attestation()	verify_credential()
pass-authenticators-ssh	verify_attestation()	verify_credential()

where c is the length of the client data (capped at 1024 bytes) and a the length of the authenticator data. Substrate keys charge the weight of the signature's key type, and Bitcoin the one of whether the key is compressed (BIP-137 flags 31-34) or not (27-30). The other authenticators' inputs have a fixed size.

The weights live in each crate's src/weights.rs, and are runtime-independent: verifying touches no storage, and the benchmarks run with their own challenger. The committed weights are placeholders, conservative estimates pending a run on reference hardware.

Running the benchmarks

Each authenticator has a benchmarking-only pallet, behind its runtime-benchmarks feature. To run them, add them to a runtime's define_benchmarks! (they don't go in construct_runtime!):

frame_benchmarking::define_benchmarks!(
    // ...
    [pass_webauthn, pass_webauthn::benchmarking::Pallet::<Runtime>]
    [pass_substrate_keys, pass_substrate_keys::benchmarking::Pallet::<Runtime>]
    [pass_ethereum, pass_ethereum::benchmarking::Pallet::<Runtime>]
    [pass_bitcoin, pass_bitcoin::benchmarking::Pallet::<Runtime>]
    [pass_nostr, pass_nostr::benchmarking::Pallet::<Runtime>]
    [pass_solana, pass_solana::benchmarking::Pallet::<Runtime>]
    [pass_ssh, pass_ssh::benchmarking::Pallet::<Runtime>]
);

Then, on reference hardware, build that runtime with --features runtime-benchmarks and regenerate the weights with this repository's template:

for pallet in webauthn substrate-keys ethereum bitcoin nostr solana ssh; do
  frame-omni-bencher v1 benchmark pallet \
    --runtime path/to/runtime.compact.compressed.wasm \
    --pallet "pass_${pallet//-/_}" --extrinsic '*' \
    --steps 50 --repeat 20 \
    --template .maintain/frame-weight-template.hbs \
    --output "authenticators/${pallet}/src/weights.rs"
done

cargo test --features runtime-benchmarks runs every benchmark once, as a test.

Benchmark helpers

With runtime-benchmarks, the attestation and credential types implement fc-traits-authn's DeviceAttestationBenchmarkHelper and CredentialBenchmarkHelper, so fc-pallet-pass's benchmarks can produce valid inputs for any runtime using these authenticators. Runtimes only need to implement ChallengerBenchmarkHelper for their challenger.