If the discrete logarithm problem turns out to be not so difficult, the entire concept of proof of work is meaningless. There is no patch or fork that can possibly fix that.
But if the DLP is easy, then anyone can fabricate proper ECDSA sigs, and you're not wasting electricity to send bogus messages, you're wasting electricity to send valid messages.
The point I was trying to make is that there is no migration path from ECDSA. You can't fork bitcoin if the DLP is easy. It's just game over. Proof of work becomes irrelevant.