Wald test Jacobian matrix

Dear Matlab, I have the following regression: mpg=b0+b1*weight+b2*foreign and I want to estimate the hypothesis that b1*b2=1. I am using the following code:

mpg=data(:,2);
weight=data(:,6);
foreign=data(:,11);
n=74;
b0=ones(n,1);
X=[b0 weight foreign];
beta_regress=regress(mpg,X)
r=[beta_regress(2,1)*beta_regress(3,1)-1];
R=[0 beta_regress(3,1) beta_regress(2,1)];
alpha=0.05;
[h,pvalue]=waldtest(r,R,alpha)

But when I run I obtain this message:

Error using waldtest (line 271)
Jacobian matrices must be q-by-p, where q is the number of
restrictions and p is the number of unrestricted parameters.
So, my question is: how should I code the Jacobian matrix?
I also have a 2nd question: how my code should change (given that I want to estimate the Wald test) in case I want to use robust standard errors?

Thanks