Is FieldVal.AddInt() in btcec really safe for overflow?

[chaintechlab]

func (f *fieldVal) AddInt(ui uint) *fieldVal {
    // Since the field representation intentionally provides overflow bits,
    // it's ok to use carryless addition as the carry bit is safely part of
    // the word and will be normalized out.
    f.n[0] += uint32(ui)
    return f
}

I think it is not really safe here. If ui is uint32::MAX and f.n[0] is 1, what will happen?
BTW, it is not proper here to cast uint to uint32 (on 64bit syste, uint is uint64).

[stevenroose]

It is documented that that function is intended for usage with small numbers:

// AddInt adds the passed integer to the existing field value and stores the
// result in f.  This is a convenience function since it is fairly common to
// perform some arithemetic with small native integers.
//
// The field value is returned to support chaining.  This enables syntax like:
// f.AddInt(1).Add(f2) so that f = f + 1 + f2.