For your information: Potential unhandled overflow in mmwavelib_log2Abs32
The following code is used this function
max = imag;
min = real;
if (real > imag)
{
max = real;
min = imag;
}
/* Approximate absolute function.
* abs (x[i] + ix[i+1]) \approx (max + min*3/8) */
absvalue = max + ((min * 3U) >> 3U);
Where imag and real are uint32_t values with 0 op to 0x80000000. There is the risk that at the "min * 3U" will perform an undetected overflow. (min >> 3U) * 3U, will almost give the same result but does not risk overflowing.
Secondly this is using very rude way of taking an absolute value, if you are taking an logarithm anyway there are more accurate ways of computing the absolute value and logarithm together with similar or better performance. Take a quick look at the function below in the same file ( mmwavelib_log2Abs16 ) and the mathematics rule: log(x^y) = y * log (x).
