Short explanation: fix how we handle the "localhost" entry in the hostfile so that the Mac (and other multi-NIC systems) will work.

Long explanation: Jeff and I spent some time chasing this down today (mostly Jeff), and found that the Mac was having problems with the replacement of "localhost" with the local nodename when we read the hostfile. Jeff then found that the Linux documentation specifically warns about the vaguery of the value returned for "nodename" (see the man page for uname for details). Sooo....when we replaced "localhost" with the local "nodename", the system couldn't figure out what node we were referring to when we tried to launch.

Solution (borrowed from LAM): if the user includes "localhost" in the hostfile, then we do NOT allow any other entries in the hostfile - the presence of another entry will generate an error message and cause mpirun to gracefully exit. Obviously, then, if "localhost" is specified in the hostfile, then we are running the application locally.

This commit was SVN r5881.

src/mca/rds/hostfile/rds_hostfile.c

@@ -34,6 +34,10 @@
 
 #include "runtime/runtime_types.h"
 
+/*
+ * Local global variable
+ */
+static bool localhost_found;
 
 static void orte_rds_hostfile_parse_error(void)
 {
@@ -76,12 +80,23 @@ static int orte_rds_hostfile_parse_line(int token, ompi_list_t* existing, ompi_l
     if (ORTE_RDS_HOSTFILE_STRING == token) {
         char* node_name = orte_rds_hostfile_value.sval;
 
-        /* convert this into something globally unique */
-        if(strcmp(node_name, "localhost") == 0) {
-            node_name = orte_system_info.nodename;
+        /* If a line for "localhost" was included, we do NOT allow
+         * any other hosts to be specified in the file. This is due to the
+         * vaguery of the "nodename" parameter returned by Linux system calls.
+         * See the man page for uname for a detailed explanation
+         */
+        if (strcmp(node_name, "localhost") == 0) {
+            if (0 < ompi_list_get_size(updates)) {
+                ORTE_ERROR_LOG(ORTE_ERR_VALUE_OUT_OF_BOUNDS);
+                return ORTE_ERR_VALUE_OUT_OF_BOUNDS;
+            }
+            localhost_found = true;
+        } else if (localhost_found) {
+                ORTE_ERROR_LOG(ORTE_ERR_VALUE_OUT_OF_BOUNDS);
+                return ORTE_ERR_VALUE_OUT_OF_BOUNDS;
         }
-
-        if(NULL == (node = orte_rds_hostfile_lookup(existing, node_name))) {
+        
+        if (NULL == (node = orte_rds_hostfile_lookup(existing, node_name))) {
             node = OBJ_NEW(orte_ras_base_node_t);
             node->node_name = strdup(node_name);
             node->node_slots = 1;
@@ -223,6 +238,9 @@ static int orte_rds_hostfile_query(void)
     ompi_list_t updates;
     ompi_list_item_t *item;
     int rc;
+    
+    /* initialize the localhost_found flag */
+    localhost_found = false;
 
     OBJ_CONSTRUCT(&existing, ompi_list_t);
     OBJ_CONSTRUCT(&updates, ompi_list_t);