Merge pull request open-mpi#7 from lrrajesh/master

klocwork, help file changes and segfault issue resolved in job launch code.

opal/class/opal_object.h

@@ -247,9 +247,11 @@ static inline opal_object_t *opal_obj_new(opal_class_t * cls);
 static inline opal_object_t *opal_obj_new_debug(opal_class_t* type, const char* file, int line)
 {
     opal_object_t* object = opal_obj_new(type);
-    object->obj_magic_id = OPAL_OBJ_MAGIC_ID;
-    object->cls_init_file_name = file;
-    object->cls_init_lineno = line;
+    if (object != NULL) {
+        object->obj_magic_id = OPAL_OBJ_MAGIC_ID;
+        object->cls_init_file_name = file;
+        object->cls_init_lineno = line;
+    }
     return object;
 }
 #define OBJ_NEW(type)                                   \

orcm/mca/sst/orcmsd/sst_orcmsd.c

@@ -146,16 +146,21 @@ static int orcmsd_init(void)
         if (ORTE_PROC_IS_HNP) {
             ORTE_PROC_MY_NAME->vpid = 0;
         } else {
-            if(NULL == mca_sst_orcmsd_component.base_vpid) {
-                ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
-                return ORTE_ERR_NOT_FOUND;
+            if (NULL == mca_sst_orcmsd_component.base_vpid) {
+                ret = ORTE_ERR_NOT_FOUND;
+                error = "requires a vpid";
+                goto error;
             }
             if (ORTE_SUCCESS != (ret = orte_util_convert_string_to_vpid(&ORTE_PROC_MY_NAME->vpid, mca_sst_orcmsd_component.base_vpid))) {
                 ORTE_ERROR_LOG(ret);
                 error = "convert_string_to_vpid";
                 goto error;
             }
         }
+    } else {
+        ret = ORTE_ERR_NOT_FOUND;
+        error = "requires a jobid";
+        goto error;
     }
 
     /* datastore - ensure we don't pickup the pmi component, but
@@ -548,8 +553,8 @@ static int orcmsd_init(void)
 
  error:
     if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
-        orte_show_help("help-orte-runtime.txt",
-                       "orte_init:startup:internal-failure",
+        orte_show_help("help-orcmsd.txt",
+                       "orcmsd_init:startup:internal-failure",
                        true, error, ORTE_ERROR_NAME(ret), ret);
     }
     return ret;
@@ -678,8 +683,8 @@ static int orcmsd_setup_node_pool(void)
     return ORTE_SUCCESS;
 
  error:
-    orte_show_help("help-orte-runtime.txt",
-                   "orte_init:startup:internal-failure",
+    orte_show_help("help-orcmsd.txt",
+                   "orcmsd_init:startup:internal-failure",
                    true, error, ORTE_ERROR_NAME(ret), ret);
 
     return ORTE_ERR_SILENT;

orcm/mca/sst/orcmsd/sst_orcmsd_component.c

@@ -18,10 +18,6 @@
 #include "orcm/mca/sst/sst.h"
 #include "orcm/mca/sst/orcmsd/sst_orcmsd.h"
 
-char *orcm_node_regex;
-char *orcm_base_jobid;
-char *orcm_base_vpid;
-
 static int component_register(void);
 static int component_open(void);
 static int component_close(void);
@@ -51,26 +47,28 @@ orcm_sst_orcmsd_component_t mca_sst_orcmsd_component = {
 
 static int component_register(void)
 {
+    int var_id;
+    mca_base_component_t *component = & mca_sst_orcmsd_component.super.base_version;
     mca_sst_orcmsd_component.node_regex = NULL;
-    (void) mca_base_var_register ("orcm", "orcm", NULL, "node_regex",
+    var_id =  mca_base_component_var_register (component, "node_regex",
                                "node_regex of allocated nodes.",
                                MCA_BASE_VAR_TYPE_STRING, NULL, 0,
                                MCA_BASE_VAR_FLAG_INTERNAL,
                                OPAL_INFO_LVL_9,
-                               MCA_BASE_VAR_SCOPE_CONSTANT,
+                               MCA_BASE_VAR_SCOPE_READONLY,
                                &mca_sst_orcmsd_component.node_regex);
 
     mca_sst_orcmsd_component.base_jobid = NULL;
-    (void) mca_base_var_register ("orcm", "orcm", "base", "jobid",
+    var_id =  mca_base_component_var_register (component, "jobid",
                                "orcmsd base jobid.",
                                MCA_BASE_VAR_TYPE_STRING, NULL, 0,
                                MCA_BASE_VAR_FLAG_INTERNAL,
                                OPAL_INFO_LVL_9,
-                               MCA_BASE_VAR_SCOPE_CONSTANT,
+                               MCA_BASE_VAR_SCOPE_READONLY,
                                &mca_sst_orcmsd_component.base_jobid);
 
     mca_sst_orcmsd_component.base_vpid = NULL;
-    (void) mca_base_var_register ("orcm", "orcm", "base", "vpid",
+    var_id =  mca_base_component_var_register (component, "vpid",
                                "orcmsd base vpid",
                                MCA_BASE_VAR_TYPE_STRING, NULL, 0,
                                MCA_BASE_VAR_FLAG_INTERNAL,

orcm/tools/orcmd/orcmd.c

@@ -403,10 +403,10 @@ slm_fork_hnp_procs(orte_jobid_t jobid, uid_t uid, gid_t gid,
         return ORTE_ERR_SYS_LIMITS_PIPES;
     }
 
-    /* find the orted binary using the install_dirs support - this also
+    /* find the orcmsd binary using the install_dirs support - this also
      * checks to ensure that we can see this executable and it *is* executable by us
      */
-    cmd = opal_path_access("orted", opal_install_dirs.bindir, X_OK);
+    cmd = opal_path_access("orcmsd", opal_install_dirs.bindir, X_OK);
     if (NULL == cmd) {
 
         /* guess we couldn't do it - best to abort */
@@ -419,7 +419,7 @@ slm_fork_hnp_procs(orte_jobid_t jobid, uid_t uid, gid_t gid,
 
     /* pass it a jobid to match my job family */
     opal_argv_append(&argc, &argv, "-mca");
-    opal_argv_append(&argc, &argv, "orcm_base_jobid");
+    opal_argv_append(&argc, &argv, "sst_orcmsd_jobid");
     if (ORTE_SUCCESS !=
         (rc = orte_util_convert_jobid_to_string(&param, jobid))) {
         ORTE_ERROR_LOG(rc);
@@ -431,7 +431,7 @@ slm_fork_hnp_procs(orte_jobid_t jobid, uid_t uid, gid_t gid,
     if( hnp ) {
         /* setup to pass the vpid */
         opal_argv_append(&argc, &argv, "-mca");
-        opal_argv_append(&argc, &argv, "orcm_base_vpid");
+        opal_argv_append(&argc, &argv, "sst_orcmsd_vpid");
         opal_argv_append(&argc, &argv, "0");
 
         /* tell the daemon it is to be the HNP */
@@ -463,7 +463,7 @@ slm_fork_hnp_procs(orte_jobid_t jobid, uid_t uid, gid_t gid,
     } else {
         /* setup to pass the vpid */
         opal_argv_append(&argc, &argv, "-mca");
-        opal_argv_append(&argc, &argv, "orcm_base_vpid");
+        opal_argv_append(&argc, &argv, "sst_orcmsd_vpid");
         opal_argv_append(&argc, &argv, "1");
         /* pass the uri of the hnp */
         asprintf(&param, "\"%s\"", hnp_uri);
@@ -479,7 +479,8 @@ slm_fork_hnp_procs(orte_jobid_t jobid, uid_t uid, gid_t gid,
 
     /* if we have static ports, pass the node list */
     if (NULL != nodes) {
-        opal_argv_append(&argc, &argv, "--nodes");
+        opal_argv_append(&argc, &argv, "-mca");
+        opal_argv_append(&argc, &argv, "sst_orcmsd_node_regex");
         opal_argv_append(&argc, &argv, nodes);
     }
 

orcm/tools/orcmsd/Makefile.am

@@ -18,6 +18,8 @@
 # $HEADER$
 #
 
+dist_orcmdata_DATA = help-orcmsd.txt
+
 include $(top_srcdir)/Makefile.ompi-rules
 
 man_pages = orcmsd.1

orcm/tools/orcmsd/help-orcmsd.txt

@@ -40,5 +40,49 @@ a core that does not exist on this node:
   node:  %s
   cores: %s
 
-The MCA param directing this behavior is orte_daemon_cores.
+The MCA param directing this behavior is orcmsd_daemon_cores.
 Please correct the request and try again.
+#
+#
+[orcmsd_init:startup:internal-failure]
+It looks like orcmsd_init failed for some reason; your parallel process is
+likely to abort.  There are many reasons that a parallel process can
+fail during orcmsd_init; some of which are due to configuration or
+environment problems.  This failure appears to be an internal failure;
+here's some additional information:
+
+  %s failed
+  --> Returned value %s (%d) instead of ORTE_SUCCESS
+#
+#
+[orcmsd:session:dir:prohibited]
+The specified location for the temporary directories required by orcmsd
+is on the list of prohibited locations:
+
+Location given: %s
+Prohibited locations: %s
+
+If you believe this is in error, please contact your system administrator
+to have the list of prohibited locations changed. Otherwise, please identify
+a different location to be used (use -h to see the cmd line option), or
+simply let the system pick a default location.
+#
+[orcmsd:session:dir:nopwname]
+orcmsd was unable to obtain the username in order to create a path
+for its required temporary directories.  This type of error is usually
+caused by a transient failure of network-based authentication services
+(e.g., LDAP or NIS failure due to network congestion), but can also be
+an indication of system misconfiguration.
+
+Please consult your system administrator about these issues and try
+again.
+#
+[orcmsd_nidmap:too_many_nodes]
+An error occurred while trying to pack the information about the job. More nodes
+have been found than the %d expected. Please check your configuration files such
+as the mapping.
+#
+[orcmsd_init:startup:num_daemons]
+orcmsd was unable to determine the number of nodes in your allocation. We
+are therefore assuming a very large number to ensure you receive proper error
+messages.

orcm/tools/orcmsd/orcmsd.1in

@@ -1,7 +1,7 @@
 .\" Man page contributed by Dirk Eddelbuettel <[email protected]>
 .\" and released under the BSD license.
 .\" Copyright (c) 2008      Sun Microsystems, Inc.  All rights reserved.
-.TH ORTED 1 "#OPAL_DATE#" "#PACKAGE_VERSION#" "#PACKAGE_NAME#"
+.TH ORCMSD 1 "#OPAL_DATE#" "#PACKAGE_VERSION#" "#PACKAGE_NAME#"
 .SH NAME
 orcmsd - Start an Open RTE User-Level Daemon
 .SH SYNOPSIS

orcm/tools/orcmsd/orcmsd.c

@@ -202,7 +202,15 @@ opal_cmd_line_init_t orcmsd_cmd_line_opts[] = {
       NULL, OPAL_CMD_LINE_TYPE_STRING,
       "Redirect output from application processes into filename.rank" },
 
-    { "orcm_node_regex", '\0', "nodes", "nodes", 1,
+    { "sst_orcmsd_node_regex", '\0', "node_regex", "node_regex", 1,
+      NULL, OPAL_CMD_LINE_TYPE_STRING,
+      "Regular expression defining nodes in system" },
+
+    { "sst_orcmsd_jobid", '\0', "jobid", "jobid", 1,
+      NULL, OPAL_CMD_LINE_TYPE_STRING,
+      "Regular expression defining nodes in system" },
+
+    { "sst_orcmsd_vpid", '\0', "vpid", "vpid", 1,
       NULL, OPAL_CMD_LINE_TYPE_STRING,
       "Regular expression defining nodes in system" },
 
@@ -301,7 +309,7 @@ int main(int argc, char *argv[])
     if (orcmsd_globals.help) {
         char *args = NULL;
         args = opal_cmd_line_get_usage_msg(cmd_line);
-        orte_show_help("help-orted.txt", "orted:usage", false,
+        orte_show_help("help-orcmsd.txt", "orted:usage", false,
                        argv[0], args);
         free(args);
         return 1;
@@ -368,7 +376,7 @@ int main(int argc, char *argv[])
                      */
                     orte_show_help_finalize();
                     /* the message will now come out locally */
-                    orte_show_help("help-orted.txt", "orted:cannot-bind",
+                    orte_show_help("help-orcmsd.txt", "orted:cannot-bind",
                                    true, orte_process_info.nodename,
                                    orte_daemon_cores);
                     ret = ORTE_ERR_NOT_SUPPORTED;

orcm/tools/orun/orun.c

@@ -734,6 +734,7 @@ int orun(int argc, char *argv[])
                 if (0 == param_len) {
                     orte_show_help("help-orun.txt", "orun:empty-prefix",
                                    true, orte_basename, orte_basename);
+                    free(param);
                     return ORTE_ERR_FATAL;
                 }
             }
@@ -1468,9 +1469,11 @@ static int create_app(int argc, char* argv[],
              */
             param = strdup(environ[i]);
             value = strchr(param, '=');
-            *value = '\0';
-            value++;
-            opal_setenv(param, value, false, &app->env);
+            if (value != NULL) {
+                *value = '\0';
+                value++;
+                opal_setenv(param, value, false, &app->env);
+            }
             free(param);
         }
     }
@@ -1537,7 +1540,7 @@ static int create_app(int argc, char* argv[],
                         free(value2);
                     }
                 } else {
-                    opal_output(0, "Warning: could not find environment variable \"%s\"\n", param);
+                    opal_output(0, "Warning: could not find environment variable \"%s\"\n", vars[i]);
                 }
             }
         }
@@ -1644,6 +1647,7 @@ static int create_app(int argc, char* argv[],
                     if (0 == param_len) {
                         orte_show_help("help-orun.txt", "orun:empty-prefix",
                                        true, orte_basename, orte_basename);
+                        free(param);
                         return ORTE_ERR_FATAL;
                     }
                 }
@@ -1992,6 +1996,7 @@ static int parse_appfile(orte_job_t *jdata, char *filename, char ***env)
             if (NULL != *env) {
                 tmp_env = opal_argv_copy(*env);
                 if (NULL == tmp_env) {
+                    fclose(fp);
                     return ORTE_ERR_OUT_OF_RESOURCE;
                 }
             } else {
@@ -2002,6 +2007,7 @@ static int parse_appfile(orte_job_t *jdata, char *filename, char ***env)
             if (ORTE_SUCCESS != rc) {
                 /* Assume that the error message has already been
                    printed; no need to cleanup -- we can just exit */
+                fclose(fp);
                 exit(1);
             }
             if (NULL != tmp_env) {