you have the null space wrong.
after the Gauss-Jordan elemination method, you have:
the columns with pivots ( elements in the diagonal ) form not the null space of A but the base of the subspace...
to know the null space you have to solve the rest of equation:
so by the matrix you have that x3 and x4 are free variables, so write x1 and x2 in function of x3 and x4 wiil give you a basis for the null space, so:

and

<=> x1 = -x2 - 2x3 + x4 and x2 = - 2x3
<=> x1 = -(-2x3) - 2x3 + x4 and x2 = - 2x3
so x1 = x4 and x2 = - 2x3
so the vectors that form the null space is:
{ ( 1, 0, 0, 1 ) , ( 0, -2, 1, 0 ) }