Use atomic variable for is_finished flag

runtime.pxd

@@ -2,7 +2,7 @@
 
 from libcpp.deque cimport deque
 from libcpp.vector cimport vector
-from libcpp.atomic cimport atomic
+from libcpp.atomic cimport atomic, memory_order_relaxed
 from libc.stdio cimport printf
 from libc.stdlib cimport rand
 from posix.unistd cimport sysconf
@@ -27,7 +27,7 @@ cdef inline void * worker_function(void * arg) nogil:
         # Wait until a queue becomes available.
         sem_wait(&sch.num_free_queues)
         # If the scheduler is finished there is nothing to do anymore.
-        if sch.is_finished:
+        if sch.is_finished.load(memory_order_relaxed):
             return <void*> 0
         # Pop or steal a queue.
         queue = worker.get_queue()
@@ -104,7 +104,7 @@ cdef cypclass Scheduler:
     sem_t num_free_queues
     atomic[int] num_pending_queues
     sem_t is_idle
-    volatile bint is_finished
+    atomic[bint] is_finished
     int num_workers
 
     lock Scheduler __new__(alloc, int num_workers=0):
@@ -113,7 +113,7 @@ cdef cypclass Scheduler:
         self.num_workers = num_workers
         sem_init(&self.num_free_queues, 0, 0)
         # Initially the scheduler is idle but not finished
-        self.is_finished = False
+        self.is_finished.store(False, memory_order_relaxed)
         sem_init(&self.is_idle, 0, 1)
         self.num_pending_queues.store(0)
         if pthread_barrier_init(&self.barrier, NULL, num_workers + 1):
@@ -162,7 +162,7 @@ cdef cypclass Scheduler:
         # Wait until all current tasks are joined
         self.join()
         # Signal the worker threads that there is no more work.
-        self.is_finished = True
+        self.is_finished.store(True, memory_order_relaxed)
         # Pretend that there are new queues to wake up the workers.
         num_free_queues = &self.num_free_queues
         for worker in self.workers: