Update ECDSA signer with bb signature once fixed

[PhilWindle]

No description provided.

[PhilWindle]

The signing code for ecdsa signatures uses a third party package as the equivalent in bb does not currently work . Once bb is fixed this should be changed.

This code is in yarn-project/circuit.js/src/barretenberg/crypto/ecdsa/index.ts.

  /**
   * Constructs an ECDSA signature given a msg and a private key.
   * @param msg - Message over which the signature is constructed.
   * @param privateKey - The secp256k1 private key of the signer.
   * @returns An ECDSA signature of the form (r, s, v).
   */
  public constructSignature(msg: Uint8Array, privateKey: Uint8Array) {
    const mem = this.wasm.call('bbmalloc', msg.length);
    this.wasm.writeMemory(0, privateKey);
    this.wasm.writeMemory(mem, msg);
    this.wasm.call('ecdsa__construct_signature', mem, msg.length, 0, 32, 64, 96);

    // TODO: Understand why this doesn't work
    // const sig = new EcdsaSignature(
    //   Buffer.from(this.wasm.getMemorySlice(32, 64)),
    //   Buffer.from(this.wasm.getMemorySlice(64, 96)),
    //   Buffer.from(this.wasm.getMemorySlice(96, 97)),
    // );

    const signature = secp256k1.sign(msg, privateKey);
    return new EcdsaSignature(
      toBufferBE(signature.r, 32),
      toBufferBE(signature.s, 32),
      numToUInt32BE(signature.recovery!).subarray(3, 4),
    );
  }

[spalladino]

The issue we ran into with touchid verification was that Noir expected signatures to be just on one side of the curve, to prevent malleability attacks. So we had to manually invert the signature before passing it. This was the code snippet, which should work the same here as long as you change the prime to the one from the secp256k1 curve. Assuming this is the issue, of course.

// Your scalar value as a BigInt
let scalar = BigInt('Your_Scalar_Value_Here');

// The order 'n' of secp256r1 as a BigInt
let n = BigInt('115792089210356248762697446949407573529996955224135760342422259061068512044369');

// Compute (n - scalar) mod n
let negated = (n - scalar) % n;

console.log(negated.toString());

Here's more info about the malleability issue.

[spalladino]

Reopening issue since barretenberg ECDSA is still not verifying in Noir. Reason why #954 passed is because the ECDSA account contract tests were temporarily commented out, and when reenabling them in #1080 signature failed.