Truely Tail-Recursive modInverse() in Prolog

Question

Rosetta code delivers me the following code snippet for modinv/3. It does calculate extended GCD via egcd/4 and then derives from it modinv/3:

egcd(_, 0, 1, 0