POE is supported on both AIX and Linux. Build POE PLM only if we find the poe binary. Fix hostfile creation and POE command line arguments.

This commit was SVN r24444.

orte/mca/plm/poe/configure.m4

@@ -11,6 +11,7 @@
 # Copyright (c) 2004-2005 The Regents of the University of California.
 #                         All rights reserved.
 # Copyright (c) 2010      Cisco Systems, Inc.  All rights reserved.
+# Copyright (c) 2010      IBM Corporation.  All rights reserved.
 # $COPYRIGHT$
 # 
 # Additional copyrights may follow
@@ -24,18 +25,22 @@
 AC_DEFUN([MCA_orte_plm_poe_CONFIG],[
     AC_CONFIG_FILES([orte/mca/plm/poe/Makefile])
 
-    # POE is only supported on AIX.  We only need executables (no
+    # POE is only supported on AIX and Linux.  We only need executables (no
     # header files or libraries), but those can be found (or not) at
-    # run-time.  So if we're on AIX, build this component.
-    AC_MSG_CHECKING([if on AIX])
+    # run-time.  So if we're on AIX or Linux, and can find the poe executable
+    # build this component.
     case $host_os in
-    aix3* | aix4* | aix5*)
-        happy=yes
+    aix* | linux* )
+        AC_CHECK_PROG(POE,poe,poe,no)
+        if test "$POE" = no ; then
+            happy=no	
+        else
+            happy=yes
+        fi
         ;;
     *)
         happy=no
         ;;
     esac
-    AC_MSG_RESULT([$happy])
     AS_IF([test "$happy" = "yes"], [$1], [$2])
 ])

orte/mca/plm/poe/plm_poe.c

@@ -12,6 +12,7 @@
  * Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
  *                         reserved. 
  * Copyright (c) 2010      Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2010      IBM Corporation.  All rights reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -190,7 +191,12 @@ static int spawn(orte_job_t *jdata)
             if (NULL == (node = (orte_node_t*)opal_pointer_array_get_item(map->nodes, i))) {
                 continue;
             }
-            fprintf(hfp,"%s\n",node->name);
+            for (j=0; j < node->procs->size; j++) {
+                if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(node->procs, j))) {
+                    continue;
+                }
+                fprintf(hfp,"%s\n",node->name);
+            }
         }
         fclose(hfp);
     }
@@ -220,14 +226,14 @@ static int spawn(orte_job_t *jdata)
                 /* vpids were assigned byslot */
                 fprintf(cfp, " OMPI_MCA_mapping=byslot");
             }
-            fprintf(cfp, " OMPI_MCA_orte_rank=%d", (int)proc->name.vpid);
-            fprintf(cfp, " OMPI_MCA_orte_num_procs=%d", (int)jdata->num_procs);
+            fprintf(cfp, " OMPI_MCA_orte_rank=%u", proc->name.vpid);
+            fprintf(cfp, " OMPI_MCA_orte_num_procs=%u", jdata->num_procs);
             fprintf(cfp, " OMPI_MCA_orte_nodes=%s", nodelist);
             fprintf(cfp, " OMPI_MCA_orte_ppn=%s", ppnlist);
-            fprintf(cfp, " OMPI_MCA_orte_app_num=%d", (int)proc->app_idx);
-            for (k=0; NULL != orte_launch_environ[k]; k++) {
+            fprintf(cfp, " OMPI_MCA_orte_app_num=%u", proc->app_idx);
+            /*for (k=0; NULL != orte_launch_environ[k]; k++) {
                 fprintf(cfp, " %s", orte_launch_environ[k]);
-            }
+            }*/
             fprintf(cfp, " %s", app->app);
             for (k=1; NULL != app->argv[k]; k++) {
                 fprintf(cfp, " %s", app->argv[k]);
@@ -263,14 +269,12 @@ static int spawn(orte_job_t *jdata)
     if(ORTE_SUCCESS!=rc) { ORTE_ERROR_LOG(rc); goto cleanup; }
     rc=poe_argv_append_int(&argv, mca_plm_poe_component.mp_retry, 0, "-retry");
     if(ORTE_SUCCESS!=rc) { ORTE_ERROR_LOG(rc); goto cleanup; }
-    rc=poe_argv_append_int(&argv, mca_plm_poe_component.mp_retrycount, 0, "-retrycount");
-    if(ORTE_SUCCESS!=rc) { ORTE_ERROR_LOG(rc); goto cleanup; }
     rc=poe_argv_append_int(&argv, mca_plm_poe_component.mp_infolevel, 0, "-infolevel");
     if(ORTE_SUCCESS!=rc) { ORTE_ERROR_LOG(rc); goto cleanup; }
 
-    if(mca_plm_poe_component.verbose>10) {
-        opal_output(0, "POE cmdline %s\n", opal_argv_join(argv, ' '));
-    }
+    opal_output_verbose(10, orte_plm_globals.output,
+                        "%s plm:poe: POE cmdline: %s\n",
+                        ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), opal_argv_join(argv, ' '));
 
     /* Start job with POE */
 

orte/mca/plm/poe/plm_poe.h

@@ -41,7 +41,6 @@ struct orte_plm_poe_component_t {
     orte_plm_base_component_t super;
     orte_jobid_t jobid;
     int priority;
-    int verbose;
     char* path;
     char* env;
     char** argv;

orte/mca/plm/poe/plm_poe_component.c

@@ -12,6 +12,7 @@
  * Copyright (c) 2007      Los Alamos National Security, LLC.  All rights
  *                         reserved. 
  * Copyright (c) 2010      Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2010      IBM Corporation.  All rights reserved.
  * $COPYRIGHT$
  * 
  * Additional copyrights may follow
@@ -111,9 +112,6 @@ int orte_plm_poe_component_open(void)
     mca_base_param_reg_int(c, "debug",
                            "Whether or not to enable debugging output for the poe plm component (0 or 1)",
                            false, false, 0, &mca_plm_poe_component.debug);
-    mca_base_param_reg_int(c, "verbose",
-                           "Verbose level",
-                           true, false, 0, &mca_plm_poe_component.verbose);
     mca_base_param_reg_int(c, "priority",
                            "Priority of the poe plm component",
                            false , false, 100, &mca_plm_poe_component.priority);