Fix a copy problem when using a drained message that has a NULL buffer. This often happens when draining a Barrier message which are sometimes point-to-point messages with count=0 and a NULL buffer.

This fixes a bug that can happen when checkpointing while one process is in such a routine. Previously a warning was thrown.

This commit was SVN r21080.

ompi/mca/crcp/bkmrk/crcp_bkmrk_pml.c

@@ -4081,12 +4081,18 @@ static int drain_message_copy_remove(ompi_crcp_bkmrk_pml_drain_message_ref_t *dr
         memcpy(*status, &drain_content_ref->status, sizeof(ompi_status_public_t)); 
     }
 
-    if( 0 != (ret = ompi_ddt_copy_content_same_ddt(datatype, count,
-                                                   (void*)buf, drain_content_ref->buffer) ) ) {
-        opal_output( mca_crcp_bkmrk_component.super.output_handle,
-                     "crcp:bkmrk: drain_message_copy_remove(): Datatype copy failed (%d)",
-                     ret);
-        exit_status = ret;
+    /* The buffer could be NULL - More likely when doing a count=0 type of message (e.g., Barrier) */
+    if( NULL == buf ) {
+        OPAL_OUTPUT_VERBOSE((20, mca_crcp_bkmrk_component.super.output_handle,
+                             "crcp:bkmrk: drain_message_copy_remove(): Skip copy - NULL buffer"));
+    } else {
+        if( 0 != (ret = ompi_ddt_copy_content_same_ddt(datatype, count,
+                                                       (void*)buf, drain_content_ref->buffer) ) ) {
+            opal_output( mca_crcp_bkmrk_component.super.output_handle,
+                         "crcp:bkmrk: drain_message_copy_remove(): Datatype copy failed (%d)",
+                         ret);
+            exit_status = ret;
+        }
     }
 
     /* Remove the message from the list */