orted/pmix: do handle OPAL_PMIX_TIMEOUT in pmix_server_lookup_fn()

Author: ggouaillardet

opal/mca/pmix/base/pmix_base_fns.c

@@ -3,7 +3,7 @@
  * Copyright (c) 2012-2015 Los Alamos National Security, LLC.  All rights
  *                         reserved.
  * Copyright (c) 2014-2016 Intel, Inc. All rights reserved.
- * Copyright (c) 2014-2015 Research Organization for Information Science
+ * Copyright (c) 2014-2016 Research Organization for Information Science
  *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
  *
@@ -152,16 +152,14 @@ int opal_pmix_base_exchange(opal_value_t *indat,
     info->type = OPAL_BOOL;
     info->data.flag = true;
     opal_list_append(&mlist, &info->super);
-    if (0 < timeout) {
-        /* give it a decent timeout as we don't know when
-         * the other side will publish - it doesn't
-         * have to be simultaneous */
-        info = OBJ_NEW(opal_value_t);
-        info->key = strdup(OPAL_PMIX_TIMEOUT);
-        info->type = OPAL_INT;
-        info->data.integer = timeout;
-        opal_list_append(&mlist, &info->super);
-    }
+    /* give it a decent timeout as we don't know when
+     * the other side will publish - it doesn't
+     * have to be simultaneous */
+    info = OBJ_NEW(opal_value_t);
+    info->key = strdup(OPAL_PMIX_TIMEOUT);
+    info->type = OPAL_INT;
+    info->data.integer = timeout;
+    opal_list_append(&mlist, &info->super);
 
     /* if a non-blocking version of lookup isn't
      * available, then use the blocking version */

orte/orted/pmix/pmix_server.c

@@ -16,7 +16,7 @@
  * Copyright (c) 2013-2015 Intel, Inc.  All rights reserved.
  * Copyright (c) 2014      Mellanox Technologies, Inc.
  *                         All rights reserved.
- * Copyright (c) 2014-2015 Research Organization for Information Science
+ * Copyright (c) 2014-2016 Research Organization for Information Science
  *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
  *
@@ -150,11 +150,17 @@ static void eviction_cbfunc(struct opal_hotel_t *hotel,
                             int room_num, void *occupant)
 {
     pmix_server_req_t *req = (pmix_server_req_t*)occupant;
+    bool timeout = false;
     int rc;
 
     /* decrement the request timeout */
-    req->timeout -= orte_pmix_server_globals.timeout;
-    if (0 < req->timeout) {
+    if (req->timeout > 0) {
+        req->timeout -= orte_pmix_server_globals.timeout;
+        if (0 >= req->timeout) {
+            timeout = true;
+        }
+    }
+    if (!timeout) {
         /* not done yet - check us back in */
         if (OPAL_SUCCESS == (rc = opal_hotel_checkin(&orte_pmix_server_globals.reqs, req, &req->room_num))) {
             return;

orte/orted/pmix/pmix_server_pub.c

@@ -16,7 +16,7 @@
  * Copyright (c) 2013-2015 Intel, Inc.  All rights reserved.
  * Copyright (c) 2014      Mellanox Technologies, Inc.
  *                         All rights reserved.
- * Copyright (c) 2014      Research Organization for Information Science
+ * Copyright (c) 2014-2016 Research Organization for Information Science
  *                         and Technology (RIST). All rights reserved.
  * $COPYRIGHT$
  *
@@ -156,13 +156,17 @@ int pmix_server_publish_fn(opal_process_name_t *proc,
         return rc;
     }
 
-    /* if we have items, pack those too - ignore persistence
+    /* if we have items, pack those too - ignore persistence, timeout
      * and range values */
     OPAL_LIST_FOREACH(iptr, info, opal_value_t) {
         if (0 == strcmp(iptr->key, OPAL_PMIX_RANGE) ||
             0 == strcmp(iptr->key, OPAL_PMIX_PERSISTENCE)) {
             continue;
         }
+        if (0 == strcmp(iptr->key, OPAL_PMIX_TIMEOUT)) {
+            req->timeout = iptr->data.integer;
+            continue;
+        }
         opal_output_verbose(5, orte_pmix_server_globals.output,
                             "%s publishing data %s of type %d from source %s",
                             ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), iptr->key, iptr->type,
@@ -250,11 +254,15 @@ int pmix_server_lookup_fn(opal_process_name_t *proc, char **keys,
         }
     }
 
-    /* if we have items, pack those too - ignore range value */
+    /* if we have items, pack those too - ignore range and timeout values */
     OPAL_LIST_FOREACH(iptr, info, opal_value_t) {
         if (0 == strcmp(iptr->key, OPAL_PMIX_RANGE)) {
             continue;
         }
+        if (0 == strcmp(iptr->key, OPAL_PMIX_TIMEOUT)) {
+            req->timeout = iptr->data.integer;
+            continue;
+        }
         if (OPAL_SUCCESS != (rc = opal_dss.pack(&req->msg, &iptr, 1, OPAL_VALUE))) {
             ORTE_ERROR_LOG(rc);
             OBJ_RELEASE(req);
@@ -340,11 +348,15 @@ int pmix_server_unpublish_fn(opal_process_name_t *proc, char **keys,
         }
     }
 
-    /* if we have items, pack those too - ignore range value */
+    /* if we have items, pack those too - ignore range and timeout values */
     OPAL_LIST_FOREACH(iptr, info, opal_value_t) {
         if (0 == strcmp(iptr->key, OPAL_PMIX_RANGE)) {
             continue;
         }
+        if (0 == strcmp(iptr->key, OPAL_PMIX_TIMEOUT)) {
+            req->timeout = iptr->data.integer;
+            continue;
+        }
         if (OPAL_SUCCESS != (rc = opal_dss.pack(&req->msg, &iptr, 1, OPAL_VALUE))) {
             ORTE_ERROR_LOG(rc);
             OBJ_RELEASE(req);