Isham (in his Lectures on Quantum Theory), in his initial chapters at least, gives examples of only such operators on such wavefunction state-spaces which have only nondegenerate eigenfunctions. I guess that this is due to the reason that it makes the (initial) formalism easier.
Hence for an operator $A$, with eigenvalues $a_1, a_2, \ldots$ with the corresponding (nondegenerate) eigenfunctions $u_1(x), u_2(x), \ldots$, we have $\text{Prob}(A=a_i;\psi)=|c_i|^2$ for a state $\psi = \sum_{i=1}^\infty c_i u_i$.
Question: What if there are two degenerate eigenfunctions corresponding to a single eigenvalue? What will expression of the probability of measuring that degenerate eigenvalue be like?