[sgen] Lock around worker start, not only finishing

If we don't take the finished_lock when starting workers, the workers that we started could have already finished the work and switch to non-parallel context before we are even able to signal all of them.

diff --git a/mono/sgen/sgen-workers.c b/mono/sgen/sgen-workers.c
index cc1930bcc2f443a65b4c536208051e89a119e94f..77c9413ac6c9a540283f3947a9a0a7b177968888 100644 (file)
--- a/mono/sgen/sgen-workers.c
+++ b/mono/sgen/sgen-workers.c
@@ -330,7 +330,9 @@ marker_idle_func (void *data_untyped)
                if (data->private_gray_queue.num_sections > 16 && workers_finished && worker_awakenings < active_workers_num) {
                        /* We bound the number of worker awakenings just to be sure */
                        worker_awakenings++;
+                       mono_os_mutex_lock (&finished_lock);
                        sgen_workers_ensure_awake ();
+                       mono_os_mutex_unlock (&finished_lock);
                }
        } else {
                worker_try_finish (data);
@@ -424,7 +426,13 @@ sgen_workers_start_all_workers (SgenObjectOperations *object_ops_nopar, SgenObje
        worker_awakenings = 0;
        mono_memory_write_barrier ();
 
+       /*
+        * We expect workers to start finishing only after all of them were awaken.
+        * Otherwise we might think that we have fewer workers and use wrong context.
+        */
+       mono_os_mutex_lock (&finished_lock);
        sgen_workers_ensure_awake ();
+       mono_os_mutex_unlock (&finished_lock);
 }
 
 void