In ethereumj, why the signature use loop to get v?

[woshidadashi]

In the java version of Ethereum, I see the signature code as follows:

public ECDSASignature sign(byte[] messageHash) {
    ECDSASignature sig = doSign(messageHash);
    // Now we have to work backwards to figure out the recId needed to recover the signature.
    int recId = -1;
    byte[] thisKey = this.pub.getEncoded(/* compressed */ false);
    for (int i = 0; i < 4; i++) {
        byte[] k = ECKey.recoverPubBytesFromSignature(i, sig, messageHash);
        if (k != null && Arrays.equals(k, thisKey)) {
            recId = i;
            break;
        }
    }
    if (recId == -1)
        throw new RuntimeException("Could not construct a recoverable key. This should never happen.");
    sig.v = (byte) (recId + 27);
    return sig;
}

Why this code use loop (for (int i = 0; i < 4; i++)) to get v (aka recId)? It can change the signature code and get v from the signing process. Can you please tell me? Thank you.

[netwalker2000]

I also made a great effort to understand this part too.
There can be up to 4 different points with a given "X coordinate modulo n". (2 because each X coordinate has two possible Y coordinates, and 2 because r+n may still be a valid X coordinate). That number between 0 and 3 we call the recovery id, or recid.
Anyone who has a better idea tells me too..

[mkalinin]

There are two possible recId values for SECP256K1: 0 and 1.

@woshidadashi what you're suggesting requires obtaining y coordinate corresponding to r component which is in its turn calculated with usage of SpongyCastle library. And iteration here is the only way of getting this bit for recId without modifying an underlying crypto lib code.

[woshidadashi]

I tried, this is very easy to modify BC lib.

[mkalinin]

Maybe you wanna make a PR and then we can see how intrusive the change is?

[woshidadashi]

Change the ECDSASigner.java like this, this can make a 3 times speed up following my test.

[mkalinin]

Thanks for pointing that out. But changing a crypto provider code is quite a risky step. Since speeding up signature generation does not affect processing speed I'd rather keep it as it is now.