pthreadpool: add a comment about a further optimization in pthreadpool_tevent_job_destructor()

This seems to be a really rare race, it's likely that the immediate
event will still trigger and cleanup.

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>

1 files changed, 17 insertions, 0 deletions

diff --git a/lib/pthreadpool/pthreadpool_tevent.c b/lib/pthreadpool/pthreadpool_tevent.c
index 3b502a7cc5a..94b6b9ded8f 100644
--- a/lib/pthreadpool/pthreadpool_tevent.c
+++ b/lib/pthreadpool/pthreadpool_tevent.c
@@ -492,6 +492,23 @@ static int pthreadpool_tevent_job_destructor(struct pthreadpool_tevent_job *job)
 	}
 
 	/*
+	 * TODO?: We could further improve this by adjusting
+	 * tevent_threaded_schedule_immediate_destructor()
+	 * and allow TALLOC_FREE() during its time
+	 * in the main_ev->scheduled_immediates list.
+	 *
+	 * PTHREAD_TEVENT_JOB_THREAD_FENCE(job);
+	 * if (state->needs_fence.signaled) {
+	 *       *
+	 *       * The signal function is completed
+	 *       * in future we may be allowed
+	 *       * to call TALLOC_FREE(job->im).
+	 *       *
+	 *      TALLOC_FREE(job->im);
+	 * }
+	 */
+
+	/*
 	 * pthreadpool_tevent_job_orphan() already removed
 	 * it from pool->jobs. And we don't need try
 	 * pthreadpool_cancel_job() again.