Truly preemptive behavior in a system is not always acceptable especially in safety critical systems. Cooperative scheduling (aka deferred preemption) is used to splits tasks into slots.
Interestingly enough, deferred preemption may lead to increased schedulabiliy of the system primary because the mutual exclusion is achieved via non-preemption and the task computabilty time tends to have lower values.
In differed preemption, we can split the application code into a non-preemptive blocks. The size of the block is determined by the maximum blocking time in the system.
The way that we model cooperative scheduling is that we split the task in two: C (completed part) and F (the final block of execution) such that:
When the equation converges i.e. , the final response time is given by: