If anyone is still curious, I've determined the simple way to do this using the incomplete beta function in MATLAB.
r to p-value
============
x=1-r^2;
n=number of samples;
p=betainc(x,0.5*(n-2),0.5);
F-value to p-value
==================
F=F-value;
v1=numerator degrees of freedom
v2=denominator degrees of freedom
x=v2/(v2+v1*F);
p=betainc(x,v2/2,v1/2);
The F-value transformation can be found as equation 26.6.2 in Abramowitz and Stegun. 1972. Handbook of Mathematical Functions. Dover, New York.
The other one was determined by looking at c-code and does not include fit or orthogonal parameters.
