Library prosa.results.fixed_priority.rta.comp.fully_preemptive

Require Export prosa.util.fixpoint.
Require Export prosa.results.fixed_priority.rta.fully_preemptive.

RTA for Fully Preemptive FP Scheduling with Computed Fixpoints

This module refines the general response-time bound for fully preemptive fixed-priority scheduling in prosa.results.fixed_priority.rta.fully_preemptive. Whereas the general result assumes that all fixpoints are given, i.e., that they have been found somehow without imposing any assumptions on how they have been found, here we compute all relevant fixpoints using the procedures provided in prosa.util.fixpoint.

Problem Setup and Analysis Context

To begin, we must set up the context as in prosa.results.fixed_priority.rta.fully_preemptive.

Section RTAforFullyPreemptiveFPModelwithArrivalCurves.

We assume ideal uni-processor schedules.

#[local] Existing Instance ideal.processor_state.

Consider any type of tasks characterized by WCETs ...

Context {Task : TaskType} `{TaskCost Task}.

... and any type of jobs associated with these tasks, where each job is characterized by an arrival time and an execution cost.

Context {Job : JobType} `{JobTask Job Task} `{JobArrival Job} `{JobCost Job}.

We assume that jobs and tasks are fully preemptive.

   #[local] Existing Instance fully_preemptive_job_model.
   #[local] Existing Instance fully_preemptive_task_model.
   #[local] Existing Instance fully_preemptive_rtc_threshold.

Consider any arrival sequence with consistent, non-duplicate arrivals.

Variable arr_seq : arrival_sequence Job.
Hypothesis H_valid_arrival_sequence : valid_arrival_sequence arr_seq.

Consider an arbitrary task set ts, ...

Variable ts : list Task.

... assume that all jobs come from the task set, ...

Hypothesis H_all_jobs_from_taskset : all_jobs_from_taskset arr_seq ts.

... and that the cost of a job does not exceed the task's WCET .

Hypothesis H_valid_job_cost: arrivals_have_valid_job_costs arr_seq.

Let max_arrivals be a family of valid arrival curves, i.e., for any task tsk in ts, max_arrival tsk is (1) an arrival bound for tsk, and (2) it is a monotonic function that equals 0 for the empty interval delta = 0.

  Context `{MaxArrivals Task}.
  Hypothesis H_valid_arrival_curve : valid_taskset_arrival_curve ts max_arrivals.
  Hypothesis H_is_arrival_curve : taskset_respects_max_arrivals arr_seq ts.

Let tsk be any task in ts that is to be analyzed.

Variable tsk : Task.
Hypothesis H_tsk_in_ts : tsk \in ts.

Recall that we assume sequential readiness, i.e., jobs of the same task are executed in order of their arrival.

#[local] Instance sequential_readiness : JobReady _ _ :=
sequential_ready_instance arr_seq.

Next, consider any ideal uniprocessor schedule of this arrival sequence.

  Variable sched : schedule (ideal.processor_state Job).
  Hypothesis H_sched_valid : valid_schedule sched arr_seq.
  Hypothesis H_jobs_come_from_arrival_sequence:
    jobs_come_from_arrival_sequence sched arr_seq.

Finally, we assume that the schedule is work-conserving ...

Hypothesis H_work_conserving : work_conserving arr_seq sched.

... and that it respects a reflexive and transitive fixed-priority scheduling policy FP.

  Context {FP : FP_policy Task}.
  Hypothesis H_priority_is_reflexive : reflexive_task_priorities FP.
  Hypothesis H_priority_is_transitive : transitive_task_priorities FP.
  Hypothesis H_respects_policy :
    respects_FP_policy_at_preemption_point arr_seq sched FP.

Computation of the Fixpoints

In the following, we introduce the bound on the maximum busy-window length L and the response-time bound R and assume that they have been are calculated using the tools provided by util.fixpoint.

Let h denote the horizon for the fixpoint search, i.e., an upper bound on the maximum relevant fixpoint. Practically speaking, this is the threshold at which the fixpoint search is aborted and ends in failure. Since we assume in the following that the fixpoint search terminated successfully, it implies that h was chosen to be "sufficiently large"; it may hence be safely ignored in the following.

Variable h : nat.

We let L denote the bound on the maximum busy-window length ...

Variable L : duration.

... and assume that it has been found by means of a successful fixpoint search.

Hypothesis H_L_is_fixpoint :
Some L = find_fixpoint (total_hep_request_bound_function_FP ts tsk) h.

Recall the notion of the search space of the RTA for fixed-priority scheduling.

Let is_in_search_space := is_in_search_space tsk L.

We let R denote tsk's response-time bound ...

Variable R : duration.

... and assume that it has been found by means of finding the maximum among the fixpoints for each element in the search space with regard to the following recurrence function.

  Let recurrence A F := task_request_bound_function tsk (A + ε)
                        + total_ohep_request_bound_function_FP ts tsk (A + F)
                        - A.
  Hypothesis H_R_is_max_fixpoint :
    Some R = find_max_fixpoint L is_in_search_space recurrence h.

Correctness of the Response-Time Bound

Using the lemmas from util.fixpoints, we can show that all preconditions of the general fully preemptive fixed-priority RTA theorem are met.

  Theorem uniprocessor_response_time_bound_fully_preemptive_fp :
    task_response_time_bound arr_seq sched tsk R.
  Proof.
    (* First, eliminate the trivial case of no higher- or equal-priority
       interference. *)
    case: (leqP (total_hep_request_bound_function_FP ts tsk ε) 0);
      first by rewrite leqn0 ⇒ /eqP ZERO; apply: pathological_total_hep_rbf_any_bound.
    move⇒ GT0.
    (* Second, apply the general result. *)
    eapply uniprocessor_response_time_bound_fully_preemptive_fp
      with (L := L) (ts := ts) ⇒ //.
    { apply: ffp_finds_positive_fixpoint; eauto.
      exact: total_hep_rbf_monotone. }
    { by apply: ffp_finds_fixpoint; eauto. }
    { move⇒ A.
      rewrite /bounded_pi.is_in_search_space ⇒ /andP[LT IN_SP].
      have IN_SP' : (A < L ) && is_in_search_space A
        by repeat (apply /andP; split).
      move: (fmf_is_maximum _ _ H_R_is_max_fixpoint IN_SP') ⇒ [F FIX BOUND].
      ∃ F; split ⇒ //.
      have: F = recurrence A F
        by apply: ffp_finds_fixpoint; eauto.
      by rewrite /recurrence; lia. }
  Qed.

End RTAforFullyPreemptiveFPModelwithArrivalCurves.