Commit e1426920 authored by Björn Brandenburg's avatar Björn Brandenburg

add readiness properties and facts on backlogged

add lemmas on consistency of backlogged jobs set

add notion of a non-clairvoyant readiness model
parent e236efb1
Require Export prosa.behavior.ready.
Require Export prosa.analysis.definitions.schedule_prefix.
Require Export prosa.model.preemption.parameter.
(** * Properties of Readiness Models *)
(** In this file, we define commonsense properties of readiness models. *)
Section ReadinessModelProperties.
(** For any type of jobs with costs and arrival times... *)
Context {Job : JobType} `{JobCost Job} `{JobArrival Job}.
(** ... and any kind of processor model, ... *)
Context {PState: Type} `{ProcessorState Job PState}.
(** ... consider a notion of job readiness. *)
Variable ReadinessModel : JobReady Job PState.
(** First, we define a notion of non-clairvoyance for readiness
models. Intuitively, whether a job is ready or not should depend only on
the past (i.e., prior allocation decisions and job behavior), not on
future events. Formally, we say that the [ReadinessModel] is
non-clairvoyant if a job's readiness at a given time does not vary across
schedules with identical prefixes. That is, given two schedules [sched]
and [sched'], the predicates [job_ready sched j t] and [job_ready sched'
j t] may not differ if [sched] and [sched'] are identical prior to time
[t]. *)
Definition nonclairvoyant_readiness :=
forall sched sched' j h,
identical_prefix sched sched' h ->
forall t,
t <= h ->
job_ready sched j t = job_ready sched' j t.
(** Next, we relate the readiness model to the preemption model. *)
Context `{JobPreemptable Job}.
(** In a preemption-policy-compliant schedule, nonpreemptive jobs must remain
scheduled. Further, in a valid schedule, scheduled jobs must be
ready. Consequently, in a valid preemption-policy-compliant schedule, a
nonpreemptive job must remain ready until at least the end of its
nonpreemptive section. *)
Definition valid_nonpreemptive_readiness :=
forall sched j t,
~~ job_preemptable j (service sched j t) -> job_ready sched j t.
End ReadinessModelProperties.
Require Export prosa.model.schedule.work_conserving.
Require Export prosa.analysis.facts.behavior.arrivals.
Require Export prosa.analysis.definitions.readiness.
(** In this file, we establish basic facts about backlogged jobs. *)
Section BackloggedJobs.
(** Consider any kind of jobs with arrival times and costs... *)
Context {Job : JobType} `{JobCost Job} `{JobArrival Job}.
(** ... and any kind of processor model, ... *)
Context {PState : Type} `{ProcessorState Job PState}.
(** ... and allow for any notion of job readiness. *)
Context `{JobReady Job PState}.
(** Given an arrival sequence and a schedule ... *)
Variable arr_seq : arrival_sequence Job.
Variable sched : schedule PState.
(** ... with consistent arrival times, ... *)
Hypothesis H_consistent_arrival_times: consistent_arrival_times arr_seq.
(** ... we observe that any backlogged job is indeed in the set of backlogged
jobs. *)
Lemma mem_backlogged_jobs:
forall j t,
arrives_in arr_seq j ->
backlogged sched j t ->
j \in jobs_backlogged_at arr_seq sched t.
rewrite /jobs_backlogged_at /arrivals_up_to.
rewrite mem_filter.
apply /andP; split; first by exact.
apply arrived_between_implies_in_arrivals => //.
rewrite /arrived_between.
apply /andP; split => //.
rewrite ltnS -/(has_arrived _ _).
now apply (backlogged_implies_arrived sched).
(** Trivially, it is also the case that any backlogged job comes from the
respective arrival sequence. *)
Lemma backlogged_job_arrives_in:
forall j t,
j \in jobs_backlogged_at arr_seq sched t ->
arrives_in arr_seq j.
move=> j t.
rewrite /jobs_backlogged_at mem_filter => /andP [_ IN].
move: IN. rewrite /arrivals_up_to.
now apply in_arrivals_implies_arrived.
End BackloggedJobs.
(** In the following section, we make one more crucial assumption: namely, that
the readiness model is non-clairvoyant, which allows us to relate
backlogged jobs in schedules with a shared prefix. *)
Section NonClairvoyance.
(** Consider any kind of jobs with arrival times and costs... *)
Context {Job : JobType} `{JobCost Job} `{JobArrival Job}.
(** ... any kind of processor model, ... *)
Context {PState : Type} `{ProcessorState Job PState}.
(** ... and allow for any non-clairvoyant notion of job readiness. *)
Context {RM : JobReady Job PState}.
Hypothesis H_nonclairvoyant_job_readiness: nonclairvoyant_readiness RM.
(** Consider any arrival sequence ... *)
Variable arr_seq : arrival_sequence Job.
(** ... and two schedules ... *)
Variable sched sched': schedule PState.
(** ... with a shared prefix to a fixed horizon. *)
Variable h : instant.
Hypothesis H_shared_prefix: identical_prefix sched sched' h.
(** We observe that a job is backlogged at a time in the prefix in one
schedule iff it is backlogged in the other schedule due to the
non-clairvoyance of the notion of job readiness ... *)
Lemma backlogged_prefix_invariance:
forall t j,
t < h ->
backlogged sched j t = backlogged sched' j t.
move=> t j IN_PREFIX.
rewrite /backlogged.
rewrite (H_nonclairvoyant_job_readiness sched sched' j h) //.
rewrite /scheduled_at H_shared_prefix //.
now apply ltnW.
(** As a corollary, if we further know that j is not scheduled at time [h],
we can expand the previous lemma to [t <= h]. *)
Corollary backlogged_prefix_invariance':
forall t j,
~~ scheduled_at sched j t ->
~~ scheduled_at sched' j t ->
t <= h ->
backlogged sched j t = backlogged sched' j t.
move=> t j NOT_SCHED NOT_SCHED'.
rewrite leq_eqVlt => /orP [/eqP EQ | LT]; last by apply backlogged_prefix_invariance.
rewrite /backlogged.
rewrite (H_nonclairvoyant_job_readiness sched sched' j h) //;
last by rewrite EQ.
now rewrite NOT_SCHED NOT_SCHED'.
(** ... and also lift this observation to the set of all backlogged jobs at
any given time in the shared prefix. *)
Lemma backlogged_jobs_prefix_invariance:
forall t,
t < h ->
jobs_backlogged_at arr_seq sched t = jobs_backlogged_at arr_seq sched' t.
move=> t IN_PREFIX.
rewrite /jobs_backlogged_at.
apply eq_filter.
rewrite /eqfun => j.
now apply backlogged_prefix_invariance.
End NonClairvoyance.
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