I'm actually quite surprised at the initial response by the PHP Core to this vulnerability. At very least I would have thought a sensible approach would be to fail securely - so if supplied with a bad hash you return false, not true!
It looks to me like it's trying to return false on a bad hash but missed one of the three ways crypt can signal failure (returning the input unmodified).
I assume s/bit/not/. The checking function is “does the given password with the stored parameters hash to the same value as the stored hash”. Hash functions are deterministic.
So returning the original hash for a valid password is the success case.
