Merge pull request #2443 from ggouaillardet/topic/v1.10/libnbc_mt

v1.10: coll/libnbc: fix race condition with multi threaded apps

Author: rhc54

ompi/mca/coll/libnbc/coll_libnbc.h

@@ -13,6 +13,8 @@
  * Copyright (c) 2008      Cisco Systems, Inc.  All rights reserved.
  * Copyright (c) 2013      Los Alamos National Security, LLC. All rights
  *                         reserved.
+ * Copyright (c) 2016      Research Organization for Information Science
+ *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
  * 
  * Additional copyrights may follow
@@ -70,7 +72,8 @@ struct ompi_coll_libnbc_component_t {
     ompi_free_list_t requests;
     opal_list_t active_requests;
     int32_t active_comms;
-    opal_atomic_lock_t progress_lock;
+    opal_atomic_lock_t progress_lock; /* protect from recursive calls */
+    opal_mutex_t lock;                /* protect access to the active_requests list */
 };
 typedef struct ompi_coll_libnbc_component_t ompi_coll_libnbc_component_t;
 

ompi/mca/coll/libnbc/coll_libnbc_component.c

@@ -89,6 +89,7 @@ libnbc_open(void)
 
     OBJ_CONSTRUCT(&mca_coll_libnbc_component.requests, ompi_free_list_t);
     OBJ_CONSTRUCT(&mca_coll_libnbc_component.active_requests, opal_list_t);
+    OBJ_CONSTRUCT(&mca_coll_libnbc_component.lock, opal_mutex_t);
     ret = ompi_free_list_init(&mca_coll_libnbc_component.requests,
                               sizeof(ompi_coll_libnbc_request_t),
                               OBJ_CLASS(ompi_coll_libnbc_request_t),
@@ -116,6 +117,7 @@ libnbc_close(void)
 
     OBJ_DESTRUCT(&mca_coll_libnbc_component.requests);
     OBJ_DESTRUCT(&mca_coll_libnbc_component.active_requests);
+    OBJ_DESTRUCT(&mca_coll_libnbc_component.lock);
 
     return OMPI_SUCCESS;
 }
@@ -240,21 +242,30 @@ ompi_coll_libnbc_progress(void)
 {
     ompi_coll_libnbc_request_t* request, *next;
 
+    /* return if invoked recursively */
     if (opal_atomic_trylock(&mca_coll_libnbc_component.progress_lock)) return 0;
 
+    /* process active requests, and use mca_coll_libnbc_component.lock to access the
+     * mca_coll_libnbc_component.active_requests list */
+    OPAL_THREAD_LOCK(&mca_coll_libnbc_component.lock);
     OPAL_LIST_FOREACH_SAFE(request, next, &mca_coll_libnbc_component.active_requests,
                            ompi_coll_libnbc_request_t) {
+        OPAL_THREAD_UNLOCK(&mca_coll_libnbc_component.lock);
         if (NBC_OK == NBC_Progress(request)) {
             /* done, remove and complete */
+            OPAL_THREAD_LOCK(&mca_coll_libnbc_component.lock);
             opal_list_remove_item(&mca_coll_libnbc_component.active_requests,
                                   &request->super.super.super);
+            OPAL_THREAD_UNLOCK(&mca_coll_libnbc_component.lock);
 
             request->super.req_status.MPI_ERROR = OMPI_SUCCESS;
             OPAL_THREAD_LOCK(&ompi_request_lock);
             ompi_request_complete(&request->super, true);
             OPAL_THREAD_UNLOCK(&ompi_request_lock);
         }
+        OPAL_THREAD_LOCK(&mca_coll_libnbc_component.lock);
     }
+    OPAL_THREAD_UNLOCK(&mca_coll_libnbc_component.lock);
 
     opal_atomic_unlock(&mca_coll_libnbc_component.progress_lock);
 

ompi/mca/coll/libnbc/nbc.c

@@ -659,7 +659,9 @@ int NBC_Start(NBC_Handle *handle, NBC_Schedule *schedule) {
   res = NBC_Start_round(handle);
   if((NBC_OK != res)) { printf("Error in NBC_Start_round() (%i)\n", res); return res; }
 
+  OPAL_THREAD_LOCK(&mca_coll_libnbc_component.lock);
   opal_list_append(&mca_coll_libnbc_component.active_requests, &(handle->super.super.super));
+  OPAL_THREAD_UNLOCK(&mca_coll_libnbc_component.lock);
 
   return NBC_OK;
 }