Fix bugs in mpint/tommath_ext_features

[maths644311798]

What problem does this PR solve?

Issue Number: Fixed #107

Possible side effects?

• Performance: Improve Performance

• Backward compatibility: No side effects

• Security: Secure

[zheyang0825]

LGTM

[maths644311798]

I request new reviews.
First, I figure out one more problem.

change MPINT_ENFORCE_OK(mp_mod_d(&q, small_prime_prod, &mod));
to MPINT_ENFORCE_OK(mp_mod_d(&q, small_prime_prod, &original_m));

Second, I can not explain why adding mp_prime_is_prime(p, 1, &res) helps pass the unittest steadily. But we need this to pass the unittest. I have tested some values of t of mp_prime_is_prime. It seems that a big t can not guarantee the generated p is a prime. But a small t can. I believe this is due to the inner structure of libtommath. But I don’t want to read all details of libtommath. I have tried to move

MPINT_ENFORCE_OK(mp_prime_is_prime(p, 1, &res));
if (!res) {
continue;
}

to the place above

MPINT_ENFORCE_OK(mp_prime_is_prime(p, t, &res));

But this does not work. I guess it may be due to the random source of libtommath.

Also note that pocklington criterion is deterministic: If q is prime and p passes pocklington criterion, then p must be prime. So we only need to do Lucas, Rabin, etc prime tests on q.

[maths644311798]

The failure of channel_brpc_blackbox_test in the unittest should not be a consequence of this pull request.

[anakinxc]

The failure of channel_brpc_blackbox_test in the unittest should not be a consequence of this pull request.

I have restarted the unittest

[usafchn]

great job