Problem with lscov for sparse input matrix with specified covariance vector

I am using lscov to solve a system of multiple small linear least squares problems set up as a block diagonal sparse matrix as follows:

blks = cell([1 size(X,2)]);

for i=1:size(X,2)

blks{i} = sparse([ones(size(X(:,i))) X(:,i)]);

end

A = blkdiag(blks{:});

Y = Y(:);

wY = wY(:);

[p,u] = lscov(A,Y); % this works for both normal and sparse A

%[p,u] = lscov(A,Y,wY); % this works when A is passed as full(A), fails when A is sparse, giving zero values for p

Wondering if anyone has thoughts on this?

Thanks,

Matthias