Cyber-physical systems (CPSs) include time-critical embedded systems where strict timing constraints are considered. These specific CPSs are dealing with close loop control functions, they are interfacing to sensors and actuators and have to fulfill safety critical real-time constraints. To guarantee the timing correctness, a lot of work introduced the strict-periodicity constraint. The latter is of great importance since it concerns sensors/actuators periodic tasks for which the relevancy of information they use is linked to the accuracy of their repetition. Applied to real-time control systems, the strict-periodicity constraint ensures much more stability and robustness. Therefore, the problem of scheduling real-time strictly periodic tasks has been targeted by several works in the literature but all these works were interested in nonpreemptive systems. In this work, we consider as strictly periodic and nonpreemptive only the sensory subtask and the remaining portion of the task may be preempted. This way, input transaction of every task receives the highest priority when accessing processor resource which enhances system predictability. On the other hand, the preemption allows easing schedulability analysis and, most of the time, it leads to a better processor utilization. The article proposes an optimal such task scheduling algorithm, called SPF for Strictly Periodic First, based on the Least Laxity First (LLF) scheduler and an efficient schedulability analysis. Simulation experiments with randomly generated task sets allowed showing the effectiveness of the proposed approach.