Remove the max_connections parameter from the radix component as it is confusing. Modify PMIx client init so that it simply returns the nspace/rank if called by a server - this allows the server to retrieve its assigned ID. Register the server's nspace so client-side operations can succeed

Signed-off-by: Ralph Castain <[email protected]>

Author: Ralph Castain

opal/mca/pmix/pmix3x/pmix/src/client/pmix_client.c

@@ -258,7 +258,7 @@ PMIX_EXPORT pmix_status_t PMIx_Init(pmix_proc_t *proc,
         return PMIX_ERR_BAD_PARAM;
     }
 
-    if (0 < pmix_globals.init_cntr) {
+    if (0 < pmix_globals.init_cntr || PMIX_PROC_SERVER == pmix_globals.proc_type) {
         /* since we have been called before, the nspace and
          * rank should be known. So return them here if
          * requested */

opal/mca/pmix/pmix3x/pmix3x_server_south.c

@@ -71,6 +71,28 @@ static void errreg_cbfunc (pmix_status_t status,
     *active = false;
 }
 
+static void opcbfunc(pmix_status_t status, void *cbdata)
+{
+    pmix3x_opcaddy_t *op = (pmix3x_opcaddy_t*)cbdata;
+
+    if (NULL != op->opcbfunc) {
+        op->opcbfunc(pmix3x_convert_rc(status), op->cbdata);
+    }
+    if (op->active) {
+        op->status = status;
+        op->active = false;
+    } else {
+        OBJ_RELEASE(op);
+    }
+}
+
+static void op2cbfunc(pmix_status_t status, void *cbdata)
+{
+    volatile bool *active = (volatile bool*)cbdata;
+
+    *active = false;
+}
+
 int pmix3x_server_init(opal_pmix_server_module_t *module,
                       opal_list_t *info)
 {
@@ -123,6 +145,12 @@ int pmix3x_server_init(opal_pmix_server_module_t *module,
     PMIx_Register_event_handler(NULL, 0, NULL, 0, pmix3x_event_hdlr, errreg_cbfunc, (void*)&active);
     PMIX_WAIT_FOR_COMPLETION(active);
 
+    /* as we might want to use some client-side functions, be sure
+     * to register our own nspace */
+    active = true;
+    PMIx_server_register_nspace(job->nspace, 1, NULL, 0, op2cbfunc, (void*)&active);
+    PMIX_WAIT_FOR_COMPLETION(active);
+
     return OPAL_SUCCESS;
 }
 
@@ -163,21 +191,6 @@ int pmix3x_server_gen_ppn(const char *input, char **ppn)
     return pmix3x_convert_rc(rc);
 }
 
-static void opcbfunc(pmix_status_t status, void *cbdata)
-{
-    pmix3x_opcaddy_t *op = (pmix3x_opcaddy_t*)cbdata;
-
-    if (NULL != op->opcbfunc) {
-        op->opcbfunc(pmix3x_convert_rc(status), op->cbdata);
-    }
-    if (op->active) {
-        op->status = status;
-        op->active = false;
-    } else {
-        OBJ_RELEASE(op);
-    }
-}
-
 static void _reg_nspace(int sd, short args, void *cbdata)
 {
     pmix3x_threadshift_t *cd = (pmix3x_threadshift_t*)cbdata;

orte/mca/ess/base/ess_base_std_orted.c

@@ -434,13 +434,6 @@ int orte_ess_base_orted_setup(char **hosts)
         orte_rml.set_contact_info(orte_process_info.my_hnp_uri);
     }
 
-    /* setup the PMIx server */
-    if (ORTE_SUCCESS != (ret = pmix_server_init())) {
-        ORTE_ERROR_LOG(ret);
-        error = "pmix server init";
-        goto error;
-    }
-
     /* select the errmgr */
     if (ORTE_SUCCESS != (ret = orte_errmgr_base_select())) {
         ORTE_ERROR_LOG(ret);

orte/mca/routed/radix/routed_radix.c

@@ -276,10 +276,6 @@ static orte_process_name_t get_route(orte_process_name_t *target)
     if (ORTE_PROC_MY_NAME->vpid == daemon.vpid) {
         ret = target;
         goto found;
-    } else if (orte_process_info.num_procs < mca_routed_radix_component.max_connections) {
-        /* if the job is small enough, send direct to the target's daemon */
-        ret = &daemon;
-        goto found;
     } else {
         /* search routing tree for next step to that daemon */
         for (item = opal_list_get_first(&my_children);

orte/mca/routed/radix/routed_radix.h

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2007      Los Alamos National Security, LLC.
  *                         All rights reserved.
+ * Copyright (c) 2016      Intel, Inc.  All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -20,7 +21,6 @@ BEGIN_C_DECLS
 typedef struct {
     orte_routed_component_t super;
     int radix;
-    orte_vpid_t max_connections;
 } orte_routed_radix_component_t;
 ORTE_MODULE_DECLSPEC extern orte_routed_radix_component_t mca_routed_radix_component;
 

orte/mca/routed/radix/routed_radix_component.c

@@ -60,14 +60,6 @@ static int orte_routed_radix_component_register(void)
                                            MCA_BASE_VAR_SCOPE_READONLY,
                                            &mca_routed_radix_component.radix);
 
-    mca_routed_radix_component.max_connections = -1;
-    (void) mca_base_component_var_register(c, "max_connections",
-                                           "Send direct between daemons if the number of nodes is less than this number",
-                                           MCA_BASE_VAR_TYPE_INT, NULL,0, 0,
-                                           OPAL_INFO_LVL_9,
-                                           MCA_BASE_VAR_SCOPE_READONLY,
-                                           &mca_routed_radix_component.max_connections);
-
     return ORTE_SUCCESS;
 }