Merge pull request #1801 from rhc54/topic/toolconnect

Add support for PMIx tool connections and queries.

.gitignore

@@ -450,6 +450,7 @@ orte/test/system/pmi_abort
 orte/test/system/opal_hwloc
 orte/test/system/opal_db
 orte/test/system/ulfm
+orte/test/system/pmixtool
 
 orte/tools/orte-checkpoint/orte-checkpoint
 orte/tools/orte-checkpoint/orte-checkpoint.1

opal/include/opal/constants.h

@@ -82,7 +82,8 @@ enum {
     OPAL_ERR_SERVER_NOT_AVAIL               = (OPAL_ERR_BASE - 52),
     OPAL_ERR_IN_PROCESS                     = (OPAL_ERR_BASE - 53),
     OPAL_ERR_DEBUGGER_RELEASE               = (OPAL_ERR_BASE - 54),
-    OPAL_ERR_HANDLERS_COMPLETE              = (OPAL_ERR_BASE - 55)
+    OPAL_ERR_HANDLERS_COMPLETE              = (OPAL_ERR_BASE - 55),
+    OPAL_ERR_PARTIAL_SUCCESS                = (OPAL_ERR_BASE - 56)
 };
 
 #define OPAL_ERR_MAX                (OPAL_ERR_BASE - 100)

opal/mca/pmix/ext20/configure.m4

@@ -59,6 +59,24 @@ AC_DEFUN([MCA_opal_pmix_ext20_CONFIG],[
                               [AC_MSG_RESULT([no])
                                opal_pmix_ext20_happy=no])
 
+           # if we have 2.0, then check further to see if we have
+           # the PMIx_Query_info function as that is even newer
+           AS_IF([test "$opal_pmix_ext20_happy" = "yes"],
+                 [AC_MSG_CHECKING([if external component is series 2.0])
+                  OPAL_CHECK_PACKAGE([opal_pmix_ext20],
+                                     [pmix.h],
+                                     [pmix],
+                                     [PMIx_Query_info],
+                                     [-lpmix],
+                                     [$pmix_ext_install_dir],
+                                     [$pmix_ext_install_dir/lib],
+                                     [AC_MSG_RESULT([yes])
+                                      opal_pmix_query_happy=1],
+                                     [AC_MSG_RESULT([no])
+                                     opal_pmix_query_happy=0])])
+
+           AC_DEFINE_UNQUOTED([HAVE_PMIX_QUERY_FUNCTION], [$opal_pmix_query_happy],
+                              [Whether or not the external library has the PMIx_Query_info function])
            AC_SUBST(opal_pmix_ext20_CPPFLAGS)
            AC_SUBST(opal_pmix_ext20_LDFLAGS)
            AC_SUBST(opal_pmix_ext20_LIBS)

opal/mca/pmix/ext20/pmix_ext20.c

@@ -350,6 +350,7 @@ static void _event_hdlr(int sd, short args, void *cbdata)
 void pmix20_event_hdlr(size_t evhdlr_registration_id,
                        pmix_status_t status, const pmix_proc_t *source,
                        pmix_info_t info[], size_t ninfo,
+                       pmix_info_t results[], size_t nresults,
                        pmix_event_notification_cbfunc_fn_t cbfunc,
                        void *cbdata)
 {
@@ -559,6 +560,9 @@ opal_pmix_data_range_t pmix20_convert_range(pmix_data_range_t range) {
 void pmix20_value_load(pmix_value_t *v,
                        opal_value_t *kv)
 {
+    size_t n;
+    char nspace[PMIX_MAX_NSLEN + 1];
+
     switch(kv->type) {
         case OPAL_UNDEF:
             v->type = PMIX_UNDEF;
@@ -650,6 +654,19 @@ void pmix20_value_load(pmix_value_t *v,
                 v->data.bo.size = 0;
             }
             break;
+        case OPAL_UINT32_ARRAY:
+            /* an array of 32-bit jobids */
+            v->type = PMIX_INFO_ARRAY;
+            v->data.array.size = kv->data.uint32_array.size;
+            if (0 < v->data.array.size) {
+                PMIX_INFO_CREATE(v->data.array.array, v->data.array.size);
+                for (n=0; n < v->data.array.size; n++) {
+                    v->data.array.array[n].value.type = PMIX_STRING;
+                    (void)opal_snprintf_jobid(nspace, PMIX_MAX_NSLEN, kv->data.uint32_array.data[n]);
+                    v->data.array.array[n].value.data.string = strdup(nspace);
+                }
+            }
+            break;
         default:
             /* silence warnings */
             break;
@@ -664,7 +681,7 @@ int pmix20_value_unload(opal_value_t *kv,
 
     switch(v->type) {
     case PMIX_UNDEF:
-        rc = OPAL_ERR_UNKNOWN_DATA_TYPE;
+        kv->type = OPAL_UNDEF;
         break;
     case PMIX_BOOL:
         kv->type = OPAL_BOOL;
@@ -1143,6 +1160,10 @@ static void ocadcon(pmix20_opalcaddy_t *p)
     p->spwncbfunc = NULL;
     p->cbdata = NULL;
     p->odmdxfunc = NULL;
+#if HAVE_PMIX_QUERY_FUNCTION
+    p->infocbfunc = NULL;
+    p->toolcbfunc = NULL;
+#endif
     p->ocbdata = NULL;
 }
 static void ocaddes(pmix20_opalcaddy_t *p)

opal/mca/pmix/ext20/pmix_ext20.h

@@ -130,6 +130,10 @@ typedef struct {
     pmix_modex_cbfunc_t mdxcbfunc;
     pmix_lookup_cbfunc_t lkupcbfunc;
     pmix_spawn_cbfunc_t spwncbfunc;
+#if HAVE_PMIX_QUERY_FUNCTION
+    pmix_info_cbfunc_t infocbfunc;
+    pmix_tool_connection_cbfunc_t toolcbfunc;
+#endif
     void *cbdata;
     opal_pmix_release_cbfunc_t odmdxfunc;
     void *ocbdata;
@@ -293,6 +297,7 @@ OPAL_MODULE_DECLSPEC int pmix20_server_notify_event(int status,
 OPAL_MODULE_DECLSPEC void pmix20_event_hdlr(size_t evhdlr_registration_id,
                                             pmix_status_t status, const pmix_proc_t *source,
                                             pmix_info_t info[], size_t ninfo,
+                                            pmix_info_t results[], size_t nresults,
                                             pmix_event_notification_cbfunc_fn_t cbfunc,
                                             void *cbdata);
 OPAL_MODULE_DECLSPEC pmix_status_t pmix20_convert_opalrc(int rc);

opal/mca/pmix/ext20/pmix_ext20_server_north.c

@@ -88,6 +88,17 @@
                                           pmix_data_range_t range,
                                           pmix_info_t info[], size_t ninfo,
                                           pmix_op_cbfunc_t cbfunc, void *cbdata);
+ #if HAVE_PMIX_QUERY_FUNCTION
+static pmix_status_t server_query(pmix_proc_t *proct,
+                                  pmix_info_t *info, size_t ninfo,
+                                  pmix_info_t *directives, size_t ndirs,
+                                  pmix_info_cbfunc_t cbfunc,
+                                  void *cbdata);
+static void server_tool_connection(pmix_info_t *info, size_t ninfo,
+                                   pmix_tool_connection_cbfunc_t cbfunc,
+                                   void *cbdata);
+#endif
+
   pmix_server_module_t mymodule = {
     .client_connected = server_client_connected_fn,
     .client_finalized = server_client_finalized_fn,
@@ -102,7 +113,11 @@
     .disconnect = server_disconnect_fn,
     .register_events = server_register_events,
     .deregister_events = server_deregister_events,
-    .notify_event = server_notify_event
+    .notify_event = server_notify_event,
+#if HAVE_PMIX_QUERY_FUNCTION
+    .query = server_query,
+    .tool_connected = server_tool_connection
+#endif
 };
 
 opal_pmix_server_module_t *host_module = NULL;
@@ -787,3 +802,159 @@ static pmix_status_t server_notify_event(pmix_status_t code,
 {
     return PMIX_ERR_NOT_SUPPORTED;
 }
+
+#if HAVE_PMIX_QUERY_FUNCTION
+static void _info_rel(void *cbdata)
+{
+    pmix20_opcaddy_t *pcaddy = (pmix20_opcaddy_t*)cbdata;
+
+    OBJ_RELEASE(pcaddy);
+}
+static void info_cbfunc(int status,
+                        opal_list_t *info,
+                        void *cbdata,
+                        opal_pmix_release_cbfunc_t release_fn,
+                        void *release_cbdata)
+{
+    pmix20_opalcaddy_t *opalcaddy = (pmix20_opalcaddy_t*)cbdata;
+    pmix20_opcaddy_t *pcaddy;
+    opal_value_t *kv;
+    size_t n;
+
+    pcaddy = OBJ_NEW(pmix20_opcaddy_t);
+
+    /* convert the status */
+    pcaddy->status = pmix20_convert_opalrc(status);
+
+    /* convert the list to a pmix_info_t array */
+    if (NULL != info) {
+        pcaddy->ninfo = opal_list_get_size(info);
+        if (0 < pcaddy->ninfo) {
+            PMIX_INFO_CREATE(pcaddy->info, pcaddy->ninfo);
+            n = 0;
+            OPAL_LIST_FOREACH(kv, info, opal_value_t) {
+                (void)strncpy(pcaddy->info[n].key, kv->key, PMIX_MAX_KEYLEN);
+                pmix20_value_load(&pcaddy->info[n].value, kv);
+            }
+        }
+    }
+    /* we are done with the incoming data */
+    if (NULL != release_fn) {
+        release_fn(release_cbdata);
+    }
+
+    /* provide the answer downward */
+    if (NULL != opalcaddy->infocbfunc) {
+        opalcaddy->infocbfunc(pcaddy->status, pcaddy->info, pcaddy->ninfo,
+                              opalcaddy->cbdata, _info_rel, pcaddy);
+    }
+    OBJ_RELEASE(opalcaddy);
+}
+
+static pmix_status_t server_query(pmix_proc_t *proct,
+                                  pmix_info_t *info, size_t ninfo,
+                                  pmix_info_t *directives, size_t ndirs,
+                                  pmix_info_cbfunc_t cbfunc,
+                                  void *cbdata)
+{
+    pmix20_opalcaddy_t *opalcaddy;
+    opal_process_name_t requestor;
+    int rc;
+    size_t n;
+    opal_value_t *oinfo;
+
+    if (NULL == host_module || NULL == host_module->query) {
+        return PMIX_ERR_NOT_SUPPORTED;
+    }
+
+    /* setup the caddy */
+    opalcaddy = OBJ_NEW(pmix20_opalcaddy_t);
+    opalcaddy->infocbfunc = cbfunc;
+    opalcaddy->cbdata = cbdata;
+
+    /* convert the requestor */
+    if (OPAL_SUCCESS != (rc = opal_convert_string_to_jobid(&requestor.jobid, proct->nspace))) {
+        OBJ_RELEASE(opalcaddy);
+        return pmix20_convert_opalrc(rc);
+    }
+    requestor.vpid = proct->rank;
+
+    /* convert the info */
+    for (n=0; n < ninfo; n++) {
+        oinfo = OBJ_NEW(opal_value_t);
+        opal_list_append(&opalcaddy->info, &oinfo->super);
+        oinfo->key = strdup(info[n].key);
+        if (OPAL_SUCCESS != (rc = pmix20_value_unload(oinfo, &info[n].value))) {
+            OBJ_RELEASE(opalcaddy);
+            return pmix20_convert_opalrc(rc);
+        }
+    }
+
+    /* we ignore directives for now */
+
+    /* pass the call upwards */
+    if (OPAL_SUCCESS != (rc = host_module->query(&requestor,
+                                                 &opalcaddy->info, NULL,
+                                                 info_cbfunc, opalcaddy))) {
+        OBJ_RELEASE(opalcaddy);
+    }
+
+    return pmix20_convert_opalrc(rc);
+}
+
+static void toolcbfunc(int status,
+                       opal_process_name_t proc,
+                       void *cbdata)
+{
+    pmix20_opalcaddy_t *opalcaddy = (pmix20_opalcaddy_t*)cbdata;
+    pmix_status_t rc;
+    pmix_proc_t p;
+
+
+    /* convert the status */
+    rc = pmix20_convert_opalrc(status);
+
+    /* convert the process name */
+    (void)opal_snprintf_jobid(p.nspace, PMIX_MAX_NSLEN, proc.jobid);
+    p.rank = proc.vpid;
+
+    /* pass it down */
+    if (NULL != opalcaddy->toolcbfunc) {
+        opalcaddy->toolcbfunc(rc, &p, opalcaddy->cbdata);
+    }
+    OBJ_RELEASE(opalcaddy);
+}
+
+static void server_tool_connection(pmix_info_t *info, size_t ninfo,
+                                   pmix_tool_connection_cbfunc_t cbfunc,
+                                   void *cbdata)
+{
+    pmix20_opalcaddy_t *opalcaddy;
+    size_t n;
+    opal_value_t *oinfo;
+    int rc;
+    pmix_status_t err;
+
+    /* setup the caddy */
+    opalcaddy = OBJ_NEW(pmix20_opalcaddy_t);
+    opalcaddy->toolcbfunc = cbfunc;
+    opalcaddy->cbdata = cbdata;
+
+    /* convert the info */
+    for (n=0; n < ninfo; n++) {
+        oinfo = OBJ_NEW(opal_value_t);
+        opal_list_append(&opalcaddy->info, &oinfo->super);
+        oinfo->key = strdup(info[n].key);
+        if (OPAL_SUCCESS != (rc = pmix20_value_unload(oinfo, &info[n].value))) {
+            OBJ_RELEASE(opalcaddy);
+            err = pmix20_convert_opalrc(rc);
+            if (NULL != cbfunc) {
+                cbfunc(err, NULL, cbdata);
+            }
+        }
+    }
+
+    /* pass it up */
+    host_module->tool_connected(&opalcaddy->info, toolcbfunc, opalcaddy);
+}
+#endif

opal/mca/pmix/pmix2x/configure.m4

@@ -41,12 +41,14 @@ AC_DEFUN([MCA_opal_pmix_pmix2x_CONFIG],[
            opal_pmix_pmix2x_save_LDFLAGS=$LDFLAGS
            opal_pmix_pmix2x_save_LIBS=$LIBS
 
-           opal_pmix_pmix2x_args="--enable-embedded-mode --with-pmix-symbol-prefix=opal_pmix_pmix2x_ --disable-visibility --with-libevent-header=\\\"opal/mca/event/$opal_event_base_include\\\" --with-hwloc-header=\\\"$opal_hwloc_base_include\\\""
+           opal_pmix_pmix2x_args="--without-tests-examples --with-pmix-symbol-prefix=opal_pmix_pmix2x_ --disable-visibility --enable-embedded-libevent --with-libevent-header=\\\"opal/mca/event/$opal_event_base_include\\\" --enable-embedded-hwloc --with-hwloc-header=\\\"$opal_hwloc_base_include\\\""
            AS_IF([test "$enable_debug" = "yes"],
                  [opal_pmix_pmix2x_args="--enable-debug $opal_pmix_pmix2x_args"
                   CFLAGS="$OPAL_CFLAGS_BEFORE_PICKY $OPAL_VISIBILITY_CFLAGS -g"],
                  [opal_pmix_pmix2x_args="--disable-debug $opal_pmix_pmix2x_args"
                   CFLAGS="$OPAL_CFLAGS_BEFORE_PICKY $OPAL_VISIBILITY_CFLAGS"])
+           AS_IF([test "$with_devel_headers" = "yes"], [],
+                 [opal_pmix_pmix2x_args="--enable-embedded-mode $opal_pmix_pmix2x_args"])
            CPPFLAGS="-I$OPAL_TOP_SRCDIR -I$OPAL_TOP_BUILDDIR -I$OPAL_TOP_SRCDIR/opal/include -I$OPAL_TOP_BUILDDIR/opal/include $CPPFLAGS"
 
            OPAL_CONFIG_SUBDIR([$opal_pmix_pmix2x_basedir/pmix],

opal/mca/pmix/pmix2x/pmix/Makefile.am

@@ -58,6 +58,8 @@ include src/client/Makefile.am
 include src/server/Makefile.am
 include src/sec/Makefile.am
 include src/event/Makefile.am
+include src/common/Makefile.am
+include src/tool/Makefile.am
 
 if WANT_DSTORE
 include src/sm/Makefile.am
@@ -74,6 +76,9 @@ else
 lib_LTLIBRARIES = libpmix.la
 libpmix_la_SOURCES = $(headers) $(sources)
 libpmix_la_LDFLAGS = -version-info $(libpmix_so_version)
+endif
+
+if PMIX_TESTS_EXAMPLES
 SUBDIRS = . test examples
 endif
 

opal/mca/pmix/pmix2x/pmix/VERSION

@@ -23,14 +23,14 @@ release=0
 # The only requirement is that it must be entirely printable ASCII
 # characters and have no white space.
 
-greek=a1
+greek=
 
 # If repo_rev is empty, then the repository version number will be
 # obtained during "make dist" via the "git describe --tags --always"
 # command, or with the date (if "git describe" fails) in the form of
 # "date<date>".
 
-repo_rev=gitaf7a389
+repo_rev=git4940b48
 
 # If tarball_version is not empty, it is used as the version string in
 # the tarball filename, regardless of all other versions listed in
@@ -44,7 +44,7 @@ tarball_version=
 
 # The date when this release was created
 
-date="Jun 16, 2016"
+date="Jun 29, 2016"
 
 # The shared library version of each of PMIx's public libraries.
 # These versions are maintained in accordance with the "Library