Correct version (C lang):

float round_offf(float num, int precision)
{
  long power = pow(10, precision + 1);;
  long result = num * power;

  if ((result % 10) >= 5)
    result += 10;

  return ((float)result / (float)power);
}