Skip to content

GitLab

Commit d0dfa9b4 authored by Felipe Cerqueira's avatar Felipe Cerqueira
Browse files

Clarify definitions and comments

parent f3b4f2a5
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 419 additions and 407 deletions
Makefile
View file @ d0dfa9b4
......@@ -14,7 +14,7 @@
#
# This Makefile was generated by the command line :
# coq_makefile -f _CoqProject ./util/ssromega.v ./util/seqset.v ./util/sorting.v ./util/step_function.v ./util/minmax.v ./util/powerset.v ./util/all.v ./util/ord_quantifier.v ./util/nat.v ./util/sum.v ./util/bigord.v ./util/counting.v ./util/tactics.v ./util/induction.v ./util/list.v ./util/divround.v ./util/bigcat.v ./util/fixedpoint.v ./util/notation.v ./analysis/global/jitter/bertogna_fp_comp.v ./analysis/global/jitter/interference_bound_edf.v ./analysis/global/jitter/workload_bound.v ./analysis/global/jitter/bertogna_edf_comp.v ./analysis/global/jitter/bertogna_fp_theory.v ./analysis/global/jitter/interference_bound.v ./analysis/global/jitter/interference_bound_fp.v ./analysis/global/jitter/bertogna_edf_theory.v ./analysis/global/parallel/bertogna_fp_comp.v ./analysis/global/parallel/interference_bound_edf.v ./analysis/global/parallel/workload_bound.v ./analysis/global/parallel/bertogna_edf_comp.v ./analysis/global/parallel/bertogna_fp_theory.v ./analysis/global/parallel/interference_bound.v ./analysis/global/parallel/interference_bound_fp.v ./analysis/global/parallel/bertogna_edf_theory.v ./analysis/global/basic/bertogna_fp_comp.v ./analysis/global/basic/interference_bound_edf.v ./analysis/global/basic/workload_bound.v ./analysis/global/basic/bertogna_edf_comp.v ./analysis/global/basic/bertogna_fp_theory.v ./analysis/global/basic/interference_bound.v ./analysis/global/basic/interference_bound_fp.v ./analysis/global/basic/bertogna_edf_theory.v ./analysis/apa/bertogna_fp_comp.v ./analysis/apa/interference_bound_edf.v ./analysis/apa/workload_bound.v ./analysis/apa/bertogna_edf_comp.v ./analysis/apa/bertogna_fp_theory.v ./analysis/apa/interference_bound.v ./analysis/apa/interference_bound_fp.v ./analysis/apa/bertogna_edf_theory.v ./analysis/uni/susp/dynamic/oblivious/fp_rta.v ./analysis/uni/susp/dynamic/oblivious/reduction.v ./analysis/uni/basic/workload_bound_fp.v ./analysis/uni/basic/fp_rta_comp.v ./analysis/uni/basic/fp_rta_theory.v ./model/arrival_sequence.v ./model/task.v ./model/task_arrival.v ./model/suspension.v ./model/partitioned/schedulability.v ./model/partitioned/schedule.v ./model/priority.v ./model/global/workload.v ./model/global/schedulability.v ./model/global/jitter/interference_edf.v ./model/global/jitter/interference.v ./model/global/jitter/job.v ./model/global/jitter/constrained_deadlines.v ./model/global/jitter/schedule.v ./model/global/jitter/platform.v ./model/global/response_time.v ./model/global/basic/interference_edf.v ./model/global/basic/interference.v ./model/global/basic/constrained_deadlines.v ./model/global/basic/schedule.v ./model/global/basic/platform.v ./model/job.v ./model/time.v ./model/arrival_bounds.v ./model/apa/interference_edf.v ./model/apa/interference.v ./model/apa/affinity.v ./model/apa/constrained_deadlines.v ./model/apa/platform.v ./model/uni/workload.v ./model/uni/transformation/construction.v ./model/uni/susp/suspension_intervals.v ./model/uni/susp/schedule.v ./model/uni/susp/platform.v ./model/uni/schedulability.v ./model/uni/schedule_of_task.v ./model/uni/response_time.v ./model/uni/schedule.v ./model/uni/basic/arrival_bounds.v ./model/uni/basic/busy_interval.v ./model/uni/basic/platform.v ./model/uni/service.v ./implementation/arrival_sequence.v ./implementation/task.v ./implementation/global/jitter/arrival_sequence.v ./implementation/global/jitter/task.v ./implementation/global/jitter/bertogna_edf_example.v ./implementation/global/jitter/job.v ./implementation/global/jitter/bertogna_fp_example.v ./implementation/global/jitter/schedule.v ./implementation/global/parallel/bertogna_edf_example.v ./implementation/global/parallel/bertogna_fp_example.v ./implementation/global/basic/bertogna_edf_example.v ./implementation/global/basic/bertogna_fp_example.v ./implementation/global/basic/schedule.v ./implementation/job.v ./implementation/apa/arrival_sequence.v ./implementation/apa/task.v ./implementation/apa/bertogna_edf_example.v ./implementation/apa/job.v ./implementation/apa/bertogna_fp_example.v ./implementation/apa/schedule.v ./implementation/uni/susp/dynamic/arrival_sequence.v ./implementation/uni/susp/dynamic/task.v ./implementation/uni/susp/dynamic/job.v ./implementation/uni/susp/dynamic/oblivious/fp_rta_example.v ./implementation/uni/susp/schedule.v ./implementation/uni/basic/fp_rta_example.v ./implementation/uni/basic/schedule.v -o Makefile
# coq_makefile -f _CoqProject ./util/ssromega.v ./util/seqset.v ./util/sorting.v ./util/step_function.v ./util/minmax.v ./util/powerset.v ./util/all.v ./util/ord_quantifier.v ./util/nat.v ./util/sum.v ./util/bigord.v ./util/counting.v ./util/tactics.v ./util/induction.v ./util/list.v ./util/divround.v ./util/bigcat.v ./util/fixedpoint.v ./util/notation.v ./analysis/global/jitter/bertogna_fp_comp.v ./analysis/global/jitter/interference_bound_edf.v ./analysis/global/jitter/workload_bound.v ./analysis/global/jitter/bertogna_edf_comp.v ./analysis/global/jitter/bertogna_fp_theory.v ./analysis/global/jitter/interference_bound.v ./analysis/global/jitter/interference_bound_fp.v ./analysis/global/jitter/bertogna_edf_theory.v ./analysis/global/parallel/bertogna_fp_comp.v ./analysis/global/parallel/interference_bound_edf.v ./analysis/global/parallel/workload_bound.v ./analysis/global/parallel/bertogna_edf_comp.v ./analysis/global/parallel/bertogna_fp_theory.v ./analysis/global/parallel/interference_bound.v ./analysis/global/parallel/interference_bound_fp.v ./analysis/global/parallel/bertogna_edf_theory.v ./analysis/global/basic/bertogna_fp_comp.v ./analysis/global/basic/interference_bound_edf.v ./analysis/global/basic/workload_bound.v ./analysis/global/basic/bertogna_edf_comp.v ./analysis/global/basic/bertogna_fp_theory.v ./analysis/global/basic/interference_bound.v ./analysis/global/basic/interference_bound_fp.v ./analysis/global/basic/bertogna_edf_theory.v ./analysis/apa/bertogna_fp_comp.v ./analysis/apa/interference_bound_edf.v ./analysis/apa/workload_bound.v ./analysis/apa/bertogna_edf_comp.v ./analysis/apa/bertogna_fp_theory.v ./analysis/apa/interference_bound.v ./analysis/apa/interference_bound_fp.v ./analysis/apa/bertogna_edf_theory.v ./analysis/uni/susp/dynamic/oblivious/fp_rta.v ./analysis/uni/susp/dynamic/oblivious/reduction.v ./analysis/uni/basic/workload_bound_fp.v ./analysis/uni/basic/fp_rta_comp.v ./analysis/uni/basic/fp_rta_theory.v ./model/arrival_sequence.v ./model/task.v ./model/task_arrival.v ./model/suspension.v ./model/partitioned/schedulability.v ./model/partitioned/schedule.v ./model/priority.v ./model/global/workload.v ./model/global/schedulability.v ./model/global/jitter/interference_edf.v ./model/global/jitter/interference.v ./model/global/jitter/job.v ./model/global/jitter/constrained_deadlines.v ./model/global/jitter/schedule.v ./model/global/jitter/platform.v ./model/global/response_time.v ./model/global/basic/interference_edf.v ./model/global/basic/interference.v ./model/global/basic/constrained_deadlines.v ./model/global/basic/schedule.v ./model/global/basic/platform.v ./model/job.v ./model/time.v ./model/arrival_bounds.v ./model/apa/interference_edf.v ./model/apa/interference.v ./model/apa/affinity.v ./model/apa/constrained_deadlines.v ./model/apa/platform.v ./model/uni/workload.v ./model/uni/transformation/construction.v ./model/uni/susp/suspension_intervals.v ./model/uni/susp/last_execution.v ./model/uni/susp/schedule.v ./model/uni/susp/platform.v ./model/uni/schedulability.v ./model/uni/schedule_of_task.v ./model/uni/response_time.v ./model/uni/schedule.v ./model/uni/basic/arrival_bounds.v ./model/uni/basic/busy_interval.v ./model/uni/basic/platform.v ./model/uni/service.v ./implementation/arrival_sequence.v ./implementation/task.v ./implementation/global/jitter/arrival_sequence.v ./implementation/global/jitter/task.v ./implementation/global/jitter/bertogna_edf_example.v ./implementation/global/jitter/job.v ./implementation/global/jitter/bertogna_fp_example.v ./implementation/global/jitter/schedule.v ./implementation/global/parallel/bertogna_edf_example.v ./implementation/global/parallel/bertogna_fp_example.v ./implementation/global/basic/bertogna_edf_example.v ./implementation/global/basic/bertogna_fp_example.v ./implementation/global/basic/schedule.v ./implementation/job.v ./implementation/apa/arrival_sequence.v ./implementation/apa/task.v ./implementation/apa/bertogna_edf_example.v ./implementation/apa/job.v ./implementation/apa/bertogna_fp_example.v ./implementation/apa/schedule.v ./implementation/uni/susp/dynamic/arrival_sequence.v ./implementation/uni/susp/dynamic/task.v ./implementation/uni/susp/dynamic/job.v ./implementation/uni/susp/dynamic/oblivious/fp_rta_example.v ./implementation/uni/susp/schedule.v ./implementation/uni/basic/fp_rta_example.v ./implementation/uni/basic/schedule.v -o Makefile
#
.DEFAULT_GOAL := all
......@@ -182,6 +182,7 @@ VFILES:=util/ssromega.v\
model/uni/workload.v\
model/uni/transformation/construction.v\
model/uni/susp/suspension_intervals.v\
model/uni/susp/last_execution.v\
model/uni/susp/schedule.v\
model/uni/susp/platform.v\
model/uni/schedulability.v\
......
analysis/uni/susp/dynamic/oblivious/fp_rta.v
View file @ d0dfa9b4
......@@ -15,7 +15,10 @@ Module SuspensionObliviousFP.
Export ResponseTimeIterationFP ReductionToBasicSchedule.
(* In this section, we formalize the suspension-oblivious RTA
for fixed-priority tasks under the dynamic self-suspension model. *)
for fixed-priority tasks under the dynamic self-suspension model.
This is just a straightforward application of the reduction
in analysis/uni/susp/dynamic/oblivious/reduction.v with the basic
response-time analysis for uniprocessor FP scheduling. *)
Section ReductionToBasicAnalysis.
Context {SporadicTask: eqType}.
......@@ -64,48 +67,48 @@ Module SuspensionObliviousFP.
Hypothesis H_dynamic_suspensions:
dynamic_suspension_model job_cost job_task next_suspension task_suspension_bound.
(* Next, consider any suspension-aware schedule... *)
Variable sched: schedule arr_seq.
(* As part of the analysis, we are going to use task costs inflated with suspension bounds, ... *)
Let inflated_cost := inflated_task_cost task_cost task_suspension_bound.
(* ...where jobs only execute after they arrive... *)
Hypothesis H_jobs_must_arrive_to_execute: jobs_must_arrive_to_execute sched.
(* ...with the condition that they are no larger than the deadline nor the period of each task. *)
Hypothesis H_inflated_cost_le_deadline_and_period:
forall tsk,
tsk \in ts ->
inflated_cost tsk <= task_deadline tsk /\
inflated_cost tsk <= task_period tsk.
(* Now we proceed with the schedulability analysis. *)
Section MainProof.
(* ...and no longer than their execution costs. *)
Hypothesis H_completed_jobs_dont_execute: completed_jobs_dont_execute job_cost sched.
(* Consider any suspension-aware schedule... *)
Variable sched: schedule arr_seq.
(* Also assume that the schedule is work-conserving when there are non-suspended jobs, ... *)
Hypothesis H_work_conserving: work_conserving job_cost next_suspension sched.
(* ...where jobs only execute after they arrive... *)
Hypothesis H_jobs_must_arrive_to_execute: jobs_must_arrive_to_execute sched.
(* ...that the schedule respects job priority... *)
Hypothesis H_respects_priority:
respects_FP_policy job_cost job_task next_suspension sched higher_eq_priority.
(* ...and no longer than their execution costs. *)
Hypothesis H_completed_jobs_dont_execute: completed_jobs_dont_execute job_cost sched.
(* ...and that suspended jobs are not allowed to be scheduled. *)
Hypothesis H_respects_self_suspensions:
respects_self_suspensions job_cost next_suspension sched.
(* Also assume that the schedule is work-conserving when there are non-suspended jobs, ... *)
Hypothesis H_work_conserving: work_conserving job_cost next_suspension sched.
(* Now we move to the schedulability analysis. *)
Section MainProof.
(* ...that the schedule respects job priority... *)
Hypothesis H_respects_priority:
respects_FP_policy job_cost job_task next_suspension sched higher_eq_priority.
(* For simplicity, let's first define some local names. *)
Let task_is_schedulable := task_misses_no_deadline job_cost job_deadline job_task sched.
(* Now, consider the task costs inflated with suspension bounds... *)
Let inflated_cost := inflated_task_cost task_cost task_suspension_bound.
(* ...and that suspended jobs are not allowed to be scheduled. *)
Hypothesis H_respects_self_suspensions:
respects_self_suspensions job_cost next_suspension sched.
(* ...and assume that these costs are no larger than the deadline nor the period of each task. *)
Hypothesis H_inflated_cost_le_deadline_and_period:
forall tsk,
tsk \in ts ->
inflated_cost tsk <= task_deadline tsk /\
inflated_cost tsk <= task_period tsk.
(* For simplicity, let's also define some local names. *)
Let task_is_schedulable := task_misses_no_deadline job_cost job_deadline job_task sched.
(* Next, recall the response-time analysis for FP scheduling instantiated with
these inflated task costs. *)
the inflated task costs. *)
Let claimed_to_be_schedulable :=
fp_schedulable inflated_cost task_period task_deadline higher_eq_priority.
(* Finally, we prove that if the suspension-oblivious schedulability test suceeds... *)
(* Then, we prove that if this suspension-oblivious response-time analysis suceeds... *)
Hypothesis H_claimed_schedulable_by_suspension_oblivious_RTA:
claimed_to_be_schedulable ts.
......
implementation/uni/susp/dynamic/oblivious/fp_rta_example.v
View file @ d0dfa9b4
......@@ -166,7 +166,7 @@ Module ResponseTimeAnalysisFP.
- by apply RM_is_reflexive.
- by apply RM_is_transitive.
- by intros tsk_a tsk_b INa INb; apply/orP; apply leq_total.
- by apply inflated_cost_le_deadline_and_period.
- by apply scheduler_jobs_must_arrive_to_execute.
- by apply scheduler_completed_jobs_dont_execute; intro j'; specialize (VALID j'); des.
- by apply scheduler_work_conserving.
......@@ -174,7 +174,6 @@ Module ResponseTimeAnalysisFP.
-- by intros t; apply RM_is_transitive.
-- by intros _ j1 j2; apply leq_total.
- by apply scheduler_respects_self_suspensions.
- by apply inflated_cost_le_deadline_and_period.
- by apply schedulability_test_succeeds.
Qed.
......
model/uni/susp/last_execution.v 0 → 100644
View file @ d0dfa9b4
Require Import rt.util.all.
Require Import rt.model.job rt.model.arrival_sequence.
Require Import rt.model.uni.schedule.
From mathcomp Require Import ssreflect ssrbool eqtype ssrnat seq fintype bigop.
(* In this file, we show how to compute the time instant after the last
execution of a job and prove several lemmas about that instant. This
notion is crucial for defining suspension intervals. *)
Module LastExecution.
Export Job UniprocessorSchedule.
(* In this section we define the time after the last execution of a job (if exists). *)
Section TimeAfterLastExecution.
Context {Job: eqType}.
Variable job_cost: Job -> time.
(* Consider any uniprocessor schedule. *)
Context {arr_seq: arrival_sequence Job}.
Variable sched: schedule arr_seq.
(* For simplicity, let's define some local names. *)
Let job_scheduled_at := scheduled_at sched.
Let job_completed_by := completed_by job_cost sched.
Section Defs.
(* Let j be any job in the arrival sequence. *)
Variable j: JobIn arr_seq.
(* Next, we will show how to find the time after the most recent
execution of a given job j in the interval [job_arrival j, t).
(Note that this instant can be time t itself.) *)
Variable t: time.
(* Let scheduled_before denote whether job j was scheduled in the interval [0, t). *)
Let scheduled_before :=
[exists t0: 'I_t, job_scheduled_at j t0].
(* In case j was scheduled before, we define the last time in which j was scheduled. *)
Let last_time_scheduled :=
\max_(t_last < t | job_scheduled_at j t_last) t_last.
(* Then, the time after the last execution of job j in the interval [0, t), if exists,
occurs:
(a) immediately after the last time in which job j was scheduled, or,
(b) if j was never scheduled, at the arrival time of j. *)
Definition time_after_last_execution :=
if scheduled_before then
last_time_scheduled + 1
else job_arrival j.
End Defs.
(* Next, we prove lemmas about the time after the last execution. *)
Section Lemmas.
(* Assume that jobs do not execute before they arrived. *)
Hypothesis H_jobs_must_arrive_to_execute:
jobs_must_arrive_to_execute sched.
(* Let j be any job in the arrival sequence. *)
Variable j: JobIn arr_seq.
(* In this section, we show that the time after the last execution occurs
no earlier than the arrival of the job. *)
Section JobHasArrived.
(* Then, the time following the last execution of job j in the
interval [0, t) occurs no earlier than the arrival of j. *)
Lemma last_execution_after_arrival:
forall t,
has_arrived j (time_after_last_execution j t).
Proof.
unfold time_after_last_execution, has_arrived; intros t.
case EX: [exists _, _]; last by done.
move: EX => /existsP [t0 SCHED].
apply leq_trans with (n := t0 + 1);
last by rewrite leq_add2r; apply leq_bigmax_cond.
apply leq_trans with (n := t0); last by rewrite addn1.
by apply H_jobs_must_arrive_to_execute.
Qed.
End JobHasArrived.
(* Next, we establish the monotonicity of the function. *)
Section Monotonicity.
(* Let t1 be any time no earlier than the arrival of job j. *)
Variable t1: time.
Hypothesis H_after_arrival: has_arrived j t1.
(* Then, (time_after_last_execution j) grows monotonically
after that point. *)
Lemma last_execution_monotonic:
forall t2,
t1 <= t2 ->
time_after_last_execution j t1 <= time_after_last_execution j t2.
Proof.
rename H_jobs_must_arrive_to_execute into ARR.
intros t2 LE12.
rewrite /time_after_last_execution.
case EX1: [exists _, _].
{
move: EX1 => /existsP [t0 SCHED0].
have EX2: [exists t:'I_t2, job_scheduled_at j t].
{
have LT: t0 < t2 by apply: (leq_trans _ LE12).
by apply/existsP; exists (Ordinal LT).
}
rewrite EX2 2!addn1.
set m1 := \max_(_ < t1 | _)_.
have LTm1: m1 < t2.
{
apply: (leq_trans _ LE12).
by apply bigmax_ltn_ord with (i0 := t0).
}
apply leq_ltn_trans with (n := Ordinal LTm1); first by done.
by apply leq_bigmax_cond, (bigmax_pred _ _ t0).
}
{
case EX2: [exists _, _]; last by done.
move: EX2 => /existsP [t0 SCHED0].
set m2 := \max_(_ < t2 | _)_.
rewrite addn1 ltnW // ltnS.
have SCHED2: scheduled_at sched j m2 by apply (bigmax_pred _ _ t0).
by apply ARR in SCHED2.
}
Qed.
End Monotonicity.
(* Next, we prove that the function is idempotent. *)
Section Idempotence.
(* The time after the last execution of job j is an idempotent function. *)
Lemma last_execution_idempotent:
forall t,
time_after_last_execution j (time_after_last_execution j t)
= time_after_last_execution j t.
Proof.
rename H_jobs_must_arrive_to_execute into ARR.
intros t.
rewrite {2 3}/time_after_last_execution.
case EX: [exists _,_].
{
move: EX => /existsP [t0 SCHED].
rewrite /time_after_last_execution.
set ex := [exists t0, _].
have EX': ex.
{
apply/existsP; rewrite addn1.
exists (Ordinal (ltnSn _)).
by apply bigmax_pred with (i0 := t0).
}
rewrite EX'; f_equal.
rewrite addn1; apply/eqP.
set m := \max_(_ < t | _)_.
have LT: m < m.+1 by done.
rewrite eqn_leq; apply/andP; split.
{
rewrite -ltnS; apply bigmax_ltn_ord with (i0 := Ordinal LT).
by apply bigmax_pred with (i0 := t0).
}
{
apply leq_trans with (n := Ordinal LT); first by done.
by apply leq_bigmax_cond, bigmax_pred with (i0 := t0).
}
}
{
apply negbT in EX; rewrite negb_exists in EX.
move: EX => /forallP EX.
rewrite /time_after_last_execution.
set ex := [exists _, _].
suff EX': ex = false; first by rewrite EX'.
apply negbTE; rewrite negb_exists; apply/forallP.
intros x.
apply/negP; intro SCHED.
apply ARR in SCHED.
by apply leq_ltn_trans with (p := job_arrival j) in SCHED;
first by rewrite ltnn in SCHED.
}
Qed.
End Idempotence.
(* Next, we show that time_after_last_execution is bounded by the identity function. *)
Section BoundedByIdentity.
(* Let t be any time no earlier than the arrival of j. *)
Variable t: time.
Hypothesis H_after_arrival: has_arrived j t.
(* Then, the time following the last execution of job j in the interval [0, t)
occurs no later than time t. *)
Lemma last_execution_bounded_by_identity:
time_after_last_execution j t <= t.
Proof.
unfold time_after_last_execution.
case EX: [exists _, _]; last by done.
move: EX => /existsP [t0 SCHED].
by rewrite addn1; apply bigmax_ltn_ord with (i0 := t0).
Qed.
End BoundedByIdentity.
(* In this section, we show that if the service received by a job
remains the same, the time after last execution also doesn't change. *)
Section SameLastExecution.
(* Consider any time instants t and t'... *)
Variable t t': time.
(* ...in which job j has received the same amount of service. *)
Hypothesis H_same_service: service sched j t = service sched j t'.
(* Then, we prove that the times after last execution relative to
instants t and t' are exactly the same. *)
Lemma same_service_implies_same_last_execution:
time_after_last_execution j t = time_after_last_execution j t'.
Proof.
rename H_same_service into SERV.
have IFF := same_service_implies_scheduled_at_earlier_times
sched j t t' SERV.
rewrite /time_after_last_execution.
rewrite IFF; case EX2: [exists _, _]; [f_equal | by done].
have EX1: [exists x: 'I_t, job_scheduled_at j x] by rewrite IFF.
clear IFF.
move: t t' SERV EX1 EX2 => t1 t2; clear t t'.
wlog: t1 t2 / t1 <= t2 => [EQ SERV EX1 EX2 | LE].
by case/orP: (leq_total t1 t2); ins; [|symmetry]; apply EQ.
set m1 := \max_(t < t1 | job_scheduled_at j t) t.
set m2 := \max_(t < t2 | job_scheduled_at j t) t.
move => SERV /existsP [t1' SCHED1'] /existsP [t2' SCHED2'].
apply/eqP; rewrite eqn_leq; apply/andP; split.
{
have WID := big_ord_widen_cond t2
(fun x => job_scheduled_at j x) (fun x => x).
rewrite /m1 /m2 {}WID //.
rewrite big_mkcond [\max_(t < t2 | _) _]big_mkcond.
apply leq_big_max; intros i _.
case AND: (_ && _); last by done.
by move: AND => /andP [SCHED _]; rewrite SCHED.
}
{
destruct (leqP t2 m1) as [GEm1 | LTm1].
{
apply leq_trans with (n := t2); last by done.
by apply ltnW, bigmax_ltn_ord with (i0 := t2').
}
destruct (ltnP m2 t1) as [LTm2 | GEm2].
{
apply leq_trans with (n := Ordinal LTm2); first by done.
by apply leq_bigmax_cond, bigmax_pred with (i0 := t2').
}
have LTm2: m2 < t2 by apply bigmax_ltn_ord with (i0 := t2').
have SCHEDm2: job_scheduled_at j m2 by apply bigmax_pred with (i0 := t2').
exfalso; move: SERV => /eqP SERV.
rewrite -[_ == _]negbK in SERV.
move: SERV => /negP SERV; apply SERV; clear SERV.
rewrite neq_ltn; apply/orP; left.
rewrite /service /service_during.
rewrite -> big_cat_nat with (n := m2) (p := t2);
[simpl | by done | by apply ltnW].
rewrite -addn1; apply leq_add; first by apply extend_sum.
destruct t2; first by rewrite ltn0 in LTm1.
rewrite big_nat_recl; last by done.
by rewrite /service_at -/job_scheduled_at SCHEDm2.
}
Qed.
End SameLastExecution.
(* In this section, we show that the service received by a job
does not change since the last execution. *)
Section SameService.
(* We prove that, for any time t, the service received by job j
before (time_after_last_execution j t) is the same as the service
by j before time t. *)
Lemma same_service_since_last_execution:
forall t,
service sched j (time_after_last_execution j t) = service sched j t.
Proof.
intros t; rewrite /time_after_last_execution.
case EX: [exists _, _].
{
move: EX => /existsP [t0 SCHED0].
set m := \max_(_ < _ | _) _; rewrite addn1.
have LTt: m < t by apply: (bigmax_ltn_ord _ _ t0).
rewrite leq_eqVlt in LTt.
move: LTt => /orP [/eqP EQ | LTt]; first by rewrite EQ.
rewrite {2}/service/service_during.
rewrite -> big_cat_nat with (n := m.+1);
[simpl | by done | by apply ltnW].
rewrite [X in _ + X]big_nat_cond [X in _ + X]big1 ?addn0 //.
move => i /andP [/andP [GTi LTi] _].
apply/eqP; rewrite eqb0; apply/negP; intro BUG.
have LEi: (Ordinal LTi) <= m by apply leq_bigmax_cond.
by apply (leq_ltn_trans LEi) in GTi; rewrite ltnn in GTi.
}
{
apply negbT in EX; rewrite negb_exists in EX.
move: EX => /forallP ALL.
rewrite /service /service_during.
rewrite ignore_service_before_arrival // big_geq //.
rewrite big_nat_cond big1 //; move => i /andP [/= LTi _].
by apply/eqP; rewrite eqb0; apply (ALL (Ordinal LTi)).
}
Qed.
End SameService.
(* In this section, we show that for any smaller value of service, we can
always find the last execution that corresponds to that service. *)
Section ExistsIntermediateExecution.
(* Assume that job j has completed by time t. *)
Variable t: time.
Hypothesis H_j_has_completed: completed_by job_cost sched j t.
(* Then, for any value of service less than the cost of j, ...*)
Variable s: time.
Hypothesis H_less_than_cost: s < job_cost j.
(* ...there exists a last execution where the service received
by job j equals s. *)
Lemma exists_last_execution_with_smaller_service:
exists t0,
service sched j (time_after_last_execution j t0) = s.
Proof.
have SAME := same_service_since_last_execution.
rename H_jobs_must_arrive_to_execute into ARR.
move: H_j_has_completed => /eqP COMP.
feed (exists_intermediate_point (service sched j));
first by apply service_is_a_step_function.
move => EX; feed (EX (job_arrival j) t).
{
feed (cumulative_service_implies_scheduled sched j 0 t);
first by apply leq_ltn_trans with (n := s);
last by rewrite -/(service _ _ _) COMP.
move => [t' [/= LTt SCHED]].
apply leq_trans with (n := t'); last by apply ltnW.
by apply ARR in SCHED.
}
feed (EX s).
{
apply/andP; split; last by rewrite COMP.
rewrite /service /service_during.
by rewrite ignore_service_before_arrival // big_geq.
}
move: EX => [x_mid [_ SERV]]; exists x_mid.
by rewrite -SERV SAME.
Qed.
End ExistsIntermediateExecution.
End Lemmas.
End TimeAfterLastExecution.
End LastExecution.
\ No newline at end of file
model/uni/susp/suspension_intervals.v
View file @ d0dfa9b4
This diff is collapsed.
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment