DBL_DIG -- maximum number of decimal digits that can be faithfully represented in a float
sys/float_info.dig
float_info.dig