Merge pull request #4206 from rhc54/topic/prun

Add a new launcher "prun" for starting applications against the ORTE DVM.
2017-09-13 06:55:30 -07:00 · 2017-09-13 06:55:30 -07:00 · df4bd83fcb
--- a/.gitignore
+++ b/.gitignore
@ -520,6 +520,8 @@ orte/tools/orted/orted
 orte/tools/orted/orted.1
 orte/tools/orterun/orterun
 orte/tools/orterun/orterun.1
+orte/tools/prun/prun
+orte/tools/prun/prun.1
 orte/tools/wrappers/ortecc-wrapper-data.txt
 orte/tools/wrappers/ortec++-wrapper-data.txt
 orte/tools/wrappers/ortecc.1
--- a/config/opal_check_pmi.m4
+++ b/config/opal_check_pmi.m4
@ -13,7 +13,7 @@
 # Copyright (c) 2009-2015 Cisco Systems, Inc.  All rights reserved.
 # Copyright (c) 2011-2014 Los Alamos National Security, LLC. All rights
 #                         reserved.
-# Copyright (c) 2014-2016 Intel, Inc. All rights reserved.
+# Copyright (c) 2014-2017 Intel, Inc. All rights reserved.
 # Copyright (c) 2014-2016 Research Organization for Information Science
 #                         and Technology (RIST). All rights reserved.
 # Copyright (c) 2016      IBM Corporation.  All rights reserved.
@ -240,9 +240,11 @@ AC_DEFUN([OPAL_CHECK_PMIX],[
           AC_MSG_ERROR([Cannot continue])])

    AC_MSG_CHECKING([if user requested external PMIx support($with_pmix)])
+    opal_prun_happy=no
    AS_IF([test -z "$with_pmix" || test "$with_pmix" = "yes" || test "$with_pmix" = "internal"],
          [AC_MSG_RESULT([no])
-           opal_external_pmix_happy=no],
+           opal_external_pmix_happy=no
+           opal_prun_happy=yes],

          [AC_MSG_RESULT([yes])
           # check for external pmix lib */
@ -295,7 +297,8 @@ AC_DEFUN([OPAL_CHECK_PMIX],[
                                                      ], [])],
                                    [AC_MSG_RESULT([found])
                                     opal_external_pmix_version=2x
-                                     opal_external_pmix_version_found=1],
+                                     opal_external_pmix_version_found=1
+                                     opal_prun_happy=yes],
                                    [AC_MSG_RESULT([not found])])])

           AS_IF([test "$opal_external_pmix_version_found" = "0"],
@ -326,5 +329,6 @@ AC_DEFUN([OPAL_CHECK_PMIX],[
           opal_external_pmix_LIBS=-lpmix
           opal_external_pmix_happy=yes])

+    AM_CONDITIONAL([OPAL_WANT_PRUN], [test "$opal_prun_happy" = "yes"])
    OPAL_VAR_SCOPE_POP
 ])
--- a/config/orte_config_files.m4
+++ b/config/orte_config_files.m4
@ -6,7 +6,7 @@
 #                         Corporation.  All rights reserved.
 # Copyright (c) 2011-2012 Los Alamos National Security, LLC.  All rights
 #                         reserved.
-# Copyright (c) 2015-2016 Intel, Inc. All rights reserved
+# Copyright (c) 2015-2017 Intel, Inc. All rights reserved.
 # $COPYRIGHT$
 #
 # Additional copyrights may follow
@ -34,5 +34,6 @@ AC_DEFUN([ORTE_CONFIG_FILES],[
        orte/tools/orte-info/Makefile
        orte/tools/orte-server/Makefile
        orte/tools/orte-dvm/Makefile
+        orte/tools/prun/Makefile
    ])
 ])
--- a/opal/mca/pmix/pmix2x/configure.m4
+++ b/opal/mca/pmix/pmix2x/configure.m4
@ -88,7 +88,7 @@ AC_DEFUN([MCA_opal_pmix_pmix2x_CONFIG],[
           opal_pmix_pmix2x_CPPFLAGS="-I$OPAL_TOP_BUILDDIR/$opal_pmix_pmix2x_basedir/pmix/include -I$OPAL_TOP_BUILDDIR/$opal_pmix_pmix2x_basedir/pmix -I$OPAL_TOP_SRCDIR/$opal_pmix_pmix2x_basedir/pmix/include -I$OPAL_TOP_SRCDIR/$opal_pmix_pmix2x_basedir/pmix"
           opal_pmix_pmix2x_DEPENDENCIES="$OPAL_TOP_BUILDDIR/$opal_pmix_pmix2x_basedir/pmix/src/libpmix.la"
           # and the flags for prun
-           OPAL_PMIX_CPPFLAGS="-I$opal_pmix_pmix2x_CPPFLAGS"
+           OPAL_PMIX_CPPFLAGS="$opal_pmix_pmix2x_CPPFLAGS"
           AC_SUBST(OPAL_PMIX_CPPFLAGS)
           OPAL_PMIX_LDADD=$opal_pmix_pmix2x_LIBS
           AC_SUBST(OPAL_PMIX_LDADD)
--- a/opal/mca/pmix/pmix2x/pmix/src/mca/ptl/tcp/ptl_tcp_component.c
+++ b/opal/mca/pmix/pmix2x/pmix/src/mca/ptl/tcp/ptl_tcp_component.c
@ -1240,6 +1240,7 @@ static void process_cbfunc(int sd, short args, void *cbdata)
    pmix_pending_connection_t *pnd = (pmix_pending_connection_t*)cd->cbdata;
    pmix_nspace_t *nptr;
    pmix_rank_info_t *info;
+    pmix_peer_t *peer;
    int rc;
    uint32_t u32;
    pmix_info_t ginfo;
@ -1294,10 +1295,24 @@ static void process_cbfunc(int sd, short args, void *cbdata)

    /* add this nspace to our pool */
    nptr = PMIX_NEW(pmix_nspace_t);
+    if (NULL == nptr) {
+        PMIX_ERROR_LOG(PMIX_ERR_NOMEM);
+        CLOSE_THE_SOCKET(pnd->sd);
+        PMIX_RELEASE(pnd);
+        PMIX_RELEASE(cd);
+        return;
+    }
    nptr->nspace = strdup(cd->proc.nspace);
    pmix_list_append(&pmix_server_globals.nspaces, &nptr->super);
    /* add this tool rank to the nspace */
    info = PMIX_NEW(pmix_rank_info_t);
+    if (NULL == info) {
+        PMIX_ERROR_LOG(PMIX_ERR_NOMEM);
+        CLOSE_THE_SOCKET(pnd->sd);
+        PMIX_RELEASE(pnd);
+        PMIX_RELEASE(cd);
+        return;
+    }
    info->pname.nspace = strdup(cd->proc.nspace);
    info->pname.rank = 0;
    /* need to include the uid/gid for validation */
@ -1306,7 +1321,14 @@ static void process_cbfunc(int sd, short args, void *cbdata)
    pmix_list_append(&nptr->ranks, &info->super);

    /* setup a peer object for this tool */
-    pmix_peer_t *peer = PMIX_NEW(pmix_peer_t);
+    peer = PMIX_NEW(pmix_peer_t);
+    if (NULL == peer) {
+        PMIX_ERROR_LOG(PMIX_ERR_NOMEM);
+        CLOSE_THE_SOCKET(pnd->sd);
+        PMIX_RELEASE(pnd);
+        PMIX_RELEASE(cd);
+        return;
+    }
    /* mark the peer proc type */
    peer->proc_type = PMIX_PROC_TOOL | pnd->proc_type;
    /* add in the nspace pointer */
@ -1382,11 +1404,11 @@ static void process_cbfunc(int sd, short args, void *cbdata)
    }

    /* start the events for this tool */
-    pmix_event_assign(&peer->recv_event, pmix_globals.evbase, pnd->sd,
+    pmix_event_assign(&peer->recv_event, pmix_globals.evbase, peer->sd,
                      EV_READ|EV_PERSIST, pmix_ptl_base_recv_handler, peer);
    pmix_event_add(&peer->recv_event, NULL);
    peer->recv_ev_active = true;
-    pmix_event_assign(&peer->send_event, pmix_globals.evbase, pnd->sd,
+    pmix_event_assign(&peer->send_event, pmix_globals.evbase, peer->sd,
                      EV_WRITE|EV_PERSIST, pmix_ptl_base_send_handler, peer);
    pmix_output_verbose(2, pmix_ptl_base_framework.framework_output,
                        "pmix:server tool %s:%d has connected on socket %d",
@ -1410,6 +1432,10 @@ static void cnct_cbfunc(pmix_status_t status,

    /* need to thread-shift this into our context */
    cd = PMIX_NEW(pmix_setup_caddy_t);
+    if (NULL == cd) {
+        PMIX_ERROR_LOG(PMIX_ERR_NOMEM);
+        return;
+    }
    cd->status = status;
    (void)strncpy(cd->proc.nspace, proc->nspace, PMIX_MAX_NSLEN);
    cd->cbdata = cbdata;
--- a/opal/mca/pmix/pmix2x/pmix2x.h
+++ b/opal/mca/pmix/pmix2x/pmix2x.h
@ -256,6 +256,12 @@ OPAL_MODULE_DECLSPEC int pmix2x_disconnectnb(opal_list_t *procs,
 OPAL_MODULE_DECLSPEC int pmix2x_resolve_peers(const char *nodename, opal_jobid_t jobid,
                                             opal_list_t *procs);
 OPAL_MODULE_DECLSPEC int pmix2x_resolve_nodes(opal_jobid_t jobid, char **nodelist);
+OPAL_MODULE_DECLSPEC int pmix2x_allocate(opal_pmix_alloc_directive_t directive,
+                                         opal_list_t *info,
+                                         opal_pmix_info_cbfunc_t cbfunc, void *cbdata);
+OPAL_MODULE_DECLSPEC int pmix2x_job_control(opal_list_t *targets,
+                                            opal_list_t *directives,
+                                            opal_pmix_info_cbfunc_t cbfunc, void *cbdata);

 /****  TOOL FUNCTIONS  ****/
 OPAL_MODULE_DECLSPEC int pmix2x_tool_init(opal_list_t *info);
--- a/opal/mca/pmix/pmix2x/pmix2x_server_north.c
+++ b/opal/mca/pmix/pmix2x/pmix2x_server_north.c
@ -670,6 +670,9 @@ static pmix_status_t server_spawn_fn(const pmix_proc_t *p,
        if (NULL != apps[n].env) {
            app->env = opal_argv_copy(apps[n].env);
        }
+        if (NULL != apps[n].cwd) {
+            app->cwd = strdup(apps[n].cwd);
+        }
        app->maxprocs = apps[n].maxprocs;
        for (k=0; k < apps[n].ninfo; k++) {
            oinfo = OBJ_NEW(opal_value_t);
--- a/orte/tools/Makefile.am
+++ b/orte/tools/Makefile.am
@ -13,7 +13,7 @@
 # Copyright (c) 2006-2008 Cisco Systems, Inc.  All rights reserved.
 # Copyright (c) 2011-2013 Los Alamos National Security, LLC.  All rights
 #                         reserved.
-# Copyright (c) 2014-2016 Intel, Inc.  All rights reserved.
+# Copyright (c) 2014-2017 Intel, Inc. All rights reserved.
 # $COPYRIGHT$
 #
 # Additional copyrights may follow
@ -50,5 +50,10 @@ DIST_SUBDIRS += \
        tools/orte-info \
        tools/orte-migrate \
        tools/orte-server \
-        tools/orte-dvm
+        tools/orte-dvm \
+        tools/prun

+if OPAL_WANT_PRUN
+SUBDIRS += \
+	tools/prun
+endif
--- a/orte/tools/orte-dvm/orte-dvm.c
+++ b/orte/tools/orte-dvm/orte-dvm.c
@ -100,7 +100,6 @@ static bool want_prefix_by_default = (bool) ORTE_WANT_ORTERUN_PREFIX_BY_DEFAULT;
 static struct {
    bool help;
    bool version;
-    char *report_uri;
    char *prefix;
    bool run_as_root;
    bool set_sid;
@ -116,10 +115,6 @@ static opal_cmd_line_init_t cmd_line_init[] = {
      &myglobals.version, OPAL_CMD_LINE_TYPE_BOOL,
      "Print version and exit" },

-    { NULL, '\0', "report-uri", "report-uri", 1,
-      &myglobals.report_uri, OPAL_CMD_LINE_TYPE_STRING,
-      "Printout URI on stdout [-], stderr [+], or a file [anything else]" },
-
    { NULL, '\0', "prefix", "prefix", 1,
      &myglobals.prefix, OPAL_CMD_LINE_TYPE_STRING,
      "Prefix to be used to look for ORTE executables" },
@ -183,7 +178,6 @@ int main(int argc, char *argv[])
    char *param, *value;
    orte_job_t *jdata=NULL;
    orte_app_context_t *app;
-    char *uri, *ptr;

    /* Setup and parse the command line */
    memset(&myglobals, 0, sizeof(myglobals));
@ -285,6 +279,11 @@ int main(int argc, char *argv[])
        exit(0);
    }

+    /* we should act as system-level PMIx server */
+    opal_setenv("OMPI_MCA_pmix_system_server", "1", true, &environ);
+    /* and as session-level PMIx server */
+    opal_setenv("OMPI_MCA_pmix_session_server", "1", true, &environ);
+
    /* Setup MCA params */
    orte_register_params();

@ -324,43 +323,7 @@ int main(int argc, char *argv[])
     */
    opal_finalize();

-    /* check for request to report uri */
-    orte_oob_base_get_addr(&uri);
-    if (NULL != myglobals.report_uri) {
-        FILE *fp;
-        if (0 == strcmp(myglobals.report_uri, "-")) {
-            /* if '-', then output to stdout */
-            printf("VMURI: %s\n", uri);
-        } else if (0 == strcmp(myglobals.report_uri, "+")) {
-            /* if '+', output to stderr */
-            fprintf(stderr, "VMURI: %s\n", uri);
-        } else if (0 == strncasecmp(myglobals.report_uri, "file:", strlen("file:"))) {
-            ptr = strchr(myglobals.report_uri, ':');
-            ++ptr;
-            fp = fopen(ptr, "w");
-            if (NULL == fp) {
-                orte_show_help("help-orterun.txt", "orterun:write_file", false,
-                               orte_basename, "pid", ptr);
-                exit(0);
-            }
-            fprintf(fp, "%s\n", uri);
-            fclose(fp);
-        } else {
-            fp = fopen(myglobals.report_uri, "w");
-            if (NULL == fp) {
-                orte_show_help("help-orterun.txt", "orterun:write_file", false,
-                               orte_basename, "pid", myglobals.report_uri);
-                exit(0);
-            }
-            fprintf(fp, "%s\n", uri);
-            fclose(fp);
-        }
-        free(uri);
-    } else {
-        printf("VMURI: %s\n", uri);
-    }
-
-    /* get the daemon job object - was created by ess/hnp component */
+     /* get the daemon job object - was created by ess/hnp component */
    if (NULL == (jdata = orte_get_job_data_object(ORTE_PROC_MY_NAME->jobid))) {
        orte_show_help("help-orterun.txt", "bad-job-object", true,
                       orte_basename);
@ -526,6 +489,8 @@ static void notify_requestor(int sd, short args, void *cbdata)
    orte_daemon_cmd_flag_t command;
    orte_grpcomm_signature_t *sig;

+opal_output(0, "NOTIFY JOB COMPLETE");
+
    /* notify the requestor */
    reply = OBJ_NEW(opal_buffer_t);

--- a/orte/tools/prun/Makefile.am
+++ b/orte/tools/prun/Makefile.am
@ -0,0 +1,65 @@
+#
+# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+#                         University Research and Technology
+#                         Corporation.  All rights reserved.
+# Copyright (c) 2004-2005 The University of Tennessee and The University
+#                         of Tennessee Research Foundation.  All rights
+#                         reserved.
+# Copyright (c) 2004-2009 High Performance Computing Center Stuttgart,
+#                         University of Stuttgart.  All rights reserved.
+# Copyright (c) 2004-2005 The Regents of the University of California.
+#                         All rights reserved.
+# Copyright (c) 2008-2014 Cisco Systems, Inc.  All rights reserved.
+# Copyright (c) 2008      Sun Microsystems, Inc.  All rights reserved.
+# Copyright (c) 2015-2017 Intel, Inc. All rights reserved.
+# $COPYRIGHT$
+#
+# Additional copyrights may follow
+#
+# $HEADER$
+#
+
+# This is not quite in the Automake spirit, but we have to do it.
+# Since the totalview portion of the library must be built with -g, we
+# must eliminate the CFLAGS that are passed in here by default (which
+# may already have debugging and/or optimization flags).  We use
+# post-processed forms of the CFLAGS in the library targets down
+# below.
+
+AM_CPPFLAGS = $(OPAL_PMIX_CPPFLAGS)
+
+CFLAGS = $(CFLAGS_WITHOUT_OPTFLAGS) $(DEBUGGER_CFLAGS)
+AM_LDFLAGS = $(OPAL_PMIX_LDFLAGS)
+
+include $(top_srcdir)/Makefile.ompi-rules
+
+man_pages = prun.1
+EXTRA_DIST = $(man_pages:.1=.1in)
+
+if OPAL_INSTALL_BINARIES
+
+bin_PROGRAMS = prun
+
+nodist_man_MANS = $(man_pages)
+
+# Ensure that the man pages are rebuilt if the opal_config.h file
+# changes; a "good enough" way to know if configure was run again (and
+# therefore the release date or version may have changed)
+$(nodist_man_MANS): $(top_builddir)/opal/include/opal_config.h
+
+endif # OPAL_INSTALL_BINARIES
+
+prun_SOURCES = \
+        main.c \
+        prun.c \
+        prun.h
+
+prun_LDADD = \
+	$(top_builddir)/orte/lib@ORTE_LIB_PREFIX@open-rte.la \
+	$(top_builddir)/opal/lib@OPAL_LIB_PREFIX@open-pal.la \
+	$(OPAL_PMIX_LDADD)
+
+prun_LIBS = $(OPAL_PMIX_LIBS)
+
+distclean-local:
+	rm -f $(man_pages)
--- a/orte/tools/prun/main.c
+++ b/orte/tools/prun/main.c
@ -0,0 +1,33 @@
+/***************************************************************************
+ *                                                                         *
+ *          Open MPI: Open Source High Performance Computing               *
+ *                                                                         *
+ *                   http://www.open-mpi.org/                              *
+ *                                                                         *
+ ***************************************************************************/
+
+#include "prun.h"
+
+int main(int argc, char *argv[])
+{
+    return prun(argc, argv);
+}
+
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2017      Intel, Inc. All rights reserved.
+ * $COPYRIGHT$
+ *
+ * Additional copyrights may follow
+ *
+ * $HEADER$
+ */
--- a/orte/tools/prun/prun.1in
+++ b/orte/tools/prun/prun.1in
--- a/orte/tools/prun/prun.c
+++ b/orte/tools/prun/prun.c
--- a/orte/tools/prun/prun.h
+++ b/orte/tools/prun/prun.h
@ -0,0 +1,37 @@
+/*
+ * Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana
+ *                         University Research and Technology
+ *                         Corporation.  All rights reserved.
+ * Copyright (c) 2004-2005 The University of Tennessee and The University
+ *                         of Tennessee Research Foundation.  All rights
+ *                         reserved.
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
+ *                         University of Stuttgart.  All rights reserved.
+ * Copyright (c) 2004-2005 The Regents of the University of California.
+ *                         All rights reserved.
+ * Copyright (c) 2007-2011 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2012-2013 Los Alamos National Security, LLC.
+ *                         All rights reserved
+ * Copyright (c) 2014-2017 Intel, Inc. All rights reserved.
+ * $COPYRIGHT$
+ *
+ * Additional copyrights may follow
+ *
+ * $HEADER$
+ */
+
+#ifndef PRUN_H
+#define PRUN_H
+
+#include "orte_config.h"
+
+BEGIN_C_DECLS
+
+/**
+ * Main body of prun functionality
+ */
+int prun(int argc, char *argv[]);
+
+END_C_DECLS
+
+#endif /* ORTERUN_ORTERUN_H */