Understand closed loop feedback control

I wish to understand, when do I sum the targeted value I want the system to reach.
when using a PID - I will sum the result and the target value before applying the controller

When describing a system dynamic system with a controller such as   u=-Kx  I will sum the target and the response after   multiplying the state with x. - what is the correct logic?

There is a lot of NPTEL lectures on this:

closed loop feedback control NPTEL