crypto.subtle.importKey fails to import ECDSA PKCS8 data that only has a private key

[sliftist]

Version

v18.16.0

Platform

MSYS_NT-10.0-22621

Subsystem

No response

What steps will reproduce the bug?

A private key without the public key fails to decode (taken from https://datatracker.ietf.org/doc/html/rfc8410#section-10.3)

var data = Uint8Array.from(atob("MC4CAQAwBQYDK2VwBCIEINTuctv5E1hK1bbY8fdp+K06/nwoy/HU++CXqI9EdVhC"), c => c.charCodeAt(0));
crypto.subtle.importKey("pkcs8", data, { name: "ECDSA", namedCurve: "P-256" }, false, ["sign"]);

This appears to be a valid private key, as seen here: https://lapo.it/asn1js/#MC4CAQAwBQYDK2VwBCIEINTuctv5E1hK1bbY8fdp-K06_nwoy_HU--CXqI9EdVhC .

How often does it reproduce? Is there a required condition?

Every time.

What is the expected behavior? Why is that the expected behavior?

ED25519 private keys are sufficient to sign payloads.

What do you see instead?

An error, Invalid key type.

Additional information

This might be an OpenSSL issue instead. I am not familiar enough with the code to trace down the exact source of the error. However, apparently OpenSSL has no issue generating public ED25519 keys from private keys (https://stackoverflow.com/questions/72151697/generating-public-ed25519-key-with-openssl), so it does appear that OpenSSL provides the functionality in some form or another.

[panva]

Ed25519 is not P-256. See the updated code that works.

var data = Uint8Array.from(
  atob("MC4CAQAwBQYDK2VwBCIEINTuctv5E1hK1bbY8fdp+K06/nwoy/HU++CXqI9EdVhC"),
  (c) => c.charCodeAt(0)
);
await crypto.subtle.importKey("pkcs8", data, "Ed25519", false, ["sign"]);

CryptoKey {
  type: 'private',
  extractable: false,
  algorithm: { name: 'Ed25519' },
  usages: [ 'sign' ]
}

Refs:

• WICG for defining these curves in webcrypto https://wicg.github.io/webcrypto-secure-curves/
• Soon~ish to be merged Curve25519 into the main webcrypto spec Add Curve25519 w3c/webcrypto#362

These curves are experimental in Node's implementation of WebCryptoAPI until they get merged into the main webcrypto spec and further adopted by more runtimes.

[sliftist]

@panva , thank you, you have no idea how helpful this information is! Closing, as it appears there is no bug.