gem-graph-server/oldsrc/scheduler.py

110 lines
4.8 KiB
Python

#=----------------------------------------------------------------------------=#
# Scheduler related functions #
# #
# Copyright © 2021 The Gem-graph Project #
# #
# This file is part of gem-graph. #
# #
# This program is free software: you can redistribute it and/or modify #
# it under the terms of the GNU Affero General Public License as #
# published by the Free Software Foundation, either version 3 of the #
# License, or (at your option) any later version. #
# #
# This program is distributed in the hope that it will be useful, #
# but WITHOUT ANY WARRANTY; without even the implied warranty of #
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
# GNU Affero General Public License for more details. #
# #
# You should have received a copy of the GNU Affero General Public License #
# along with this program. If not, see <https://www.gnu.org/licenses/>. #
#=----------------------------------------------------------------------------=#
import os, multiprocessing, threading
import random
import localthread
## Master Thread
class GreatScheduler(threading.Thread):
def __init__(self, preemption_space, transitions_tree, arrow_list, n_thread, nmax_cycles):
threading.Thread.__init__(self)
self.preemption_space = preemption_space
self.transition_tree = transitions_tree
self.arrow_list = arrow_list
if n_thread == 0: # No limit
# Use number of usable CPUs multiplied by 3
self.n_thread = len(os.sched_getaffinity(0)) * 3
else:
self.n_thread = n_thread
self.cur_id = -1
self.n_cycle = -1
self.nmax_cycles = nmax_cycles
self.stopped = False
def run(self):
thread_list = []
shared_memory = multiprocessing.Manager()
while not self.stopped and self.n_cycle != self.nmax_cycles:
self.n_cycle += 1
#print("--- Mesures et écoute ---")
#print("Espace global de préemption :\t {}".format(self.preemption_space))
#print("Liste des flèches :\t {}".format(self.arrow_list))
## Tirer une flêche au hasard
n = int(random.random()*len(self.arrow_list))
elected_arrow = self.arrow_list[n]
## Vérification de l'espace global de préemption autour de la flêche
if not(len(thread_list) >= self.n_thread) and self.preemption_space[elected_arrow[0]]:
print("Espace local libre à {}".format(elected_arrow[0]))
self.preemption_space[elected_arrow[0]] = False
## Création du thread local
self.cur_id += 1
thread_list.append(
localthread.LocalComputingUnit(self.cur_id,
shared_memory,
elected_arrow[0],
elected_arrow[1],
self.transition_tree
)
)
## Lancer le thread local
thread_list[-1].start()
for thread in thread_list:
if not thread.is_alive():
if not thread.namespace.returncode.__class__ is localthread.SuccessfulOperation:
if thread.namespace.returncode.__class__ is localthread.FailedOperation:
print("A optimiser : créer une règle")
else:
raise thread.namespace.returncode
print("Thread local n°{} est terminé".format(thread.id))
self.preemption_space[thread.address] = True
thread_list.pop(thread_list.index(thread))
## Sortie du scheduler
print("Attente de la fin des thread locaux")
for thread in thread_list:
thread.join()
if not thread.namespace.returncode.__class__ is localthread.SuccessfulOperation:
if thread.namespace.returncode.__class__ is localthread.FailedOperation:
print("A optimiser : créer une règle")
else:
raise thread.namespace.returncode
print("Thread local n°{} est terminé".format(thread.id))