Cleanup the sensor code:

* use the global flags for linux and apple being found instead of re-doing the case statements

* update select procedure to ignore components that measure the same thing (e.g., resusage and sigar), taking the higher priority module

cmr=v1.7.5:reviewer=jsquyres:subject=Cleanup the sensor code

This commit was SVN r30368.

orte/mca/sensor/base/sensor_base_fns.c

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2010      Cisco Systems, Inc.  All rights reserved. 
  * Copyright (c) 2012      Los Alamos National Security, Inc. All rights reserved.
+ * Copyright (c) 2014      Intel, Inc. All rights reserved.
  *
  * $COPYRIGHT$
  *
@@ -117,7 +118,7 @@ void orte_sensor_base_sample(int fd, short args, void *cbdata)
             opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                                 "%s sensor:base: sampling component %s",
                                 ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
-                                i_module->component->mca_component_name);
+                                i_module->component->base_version.mca_component_name);
             i_module->module->sample();
         }
     }
@@ -147,7 +148,7 @@ void orte_sensor_base_log(char *comp, opal_buffer_t *data)
         if (NULL == (i_module = (orte_sensor_active_module_t*)opal_pointer_array_get_item(&orte_sensor_base.modules, i))) {
             continue;
         }
-        if (0 == strcmp(comp, i_module->component->mca_component_name)) {
+        if (0 == strcmp(comp, i_module->component->base_version.mca_component_name)) {
             if (NULL != i_module->module->log) {
                 i_module->module->log(data);
             }

orte/mca/sensor/base/sensor_base_select.c

@@ -41,7 +41,7 @@ static bool selected = false;
 int orte_sensor_base_select(void)
 {
     mca_base_component_list_item_t *cli = NULL;
-    mca_base_component_t *component = NULL;
+    orte_sensor_base_component_t *component = NULL;
     mca_base_module_t *module = NULL;
     orte_sensor_active_module_t *i_module;
     int priority = 0, i, j, low_i;
@@ -49,6 +49,7 @@ int orte_sensor_base_select(void)
     bool none_found;
     orte_sensor_active_module_t *tmp_module = NULL, *tmp_module_sw = NULL;
     orte_job_t *jdata;
+    bool duplicate;
 
     if (selected) {
         return ORTE_SUCCESS;
@@ -66,15 +67,15 @@ int orte_sensor_base_select(void)
      */
     none_found = true;
     OPAL_LIST_FOREACH(cli, &orte_sensor_base_framework.framework_components, mca_base_component_list_item_t) {
-        component = (mca_base_component_t *) cli->cli_component;
+        component = (orte_sensor_base_component_t *) cli->cli_component;
 
         /*
          * If there is a query function then use it.
          */
-        if (NULL == component->mca_query_component) {
+        if (NULL == component->base_version.mca_query_component) {
             opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                                 "sensor:base:select Skipping component [%s]. It does not implement a query function",
-                                component->mca_component_name );
+                                component->base_version.mca_component_name );
             continue;
         }
 
@@ -83,9 +84,9 @@ int orte_sensor_base_select(void)
          */
         opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                             "sensor:base:select Querying component [%s]",
-                            component->mca_component_name);
+                            component->base_version.mca_component_name);
 
-        component->mca_query_component(&module, &priority);
+        component->base_version.mca_query_component(&module, &priority);
 
         /*
          * If no module was returned or negative priority, then skip component
@@ -93,7 +94,40 @@ int orte_sensor_base_select(void)
         if (NULL == module || priority < 0) {
             opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                                 "sensor:base:select Skipping component [%s]. Query failed to return a module",
-                                component->mca_component_name );
+                                component->base_version.mca_component_name );
+            continue;
+        }
+
+        /* check to see if we already have someone who senses the
+         * same things - if so, take the higher priority one
+         */
+        duplicate = false;
+        for (i=0; i < tmp_array.size; i++) {
+            tmp_module = (orte_sensor_active_module_t*)opal_pointer_array_get_item(&tmp_array, i);
+            if (NULL == tmp_module) {
+                continue;
+            }
+            if (0 == strcmp(component->data_measured, tmp_module->component->data_measured)) {
+                if (tmp_module->priority < priority) {
+                    opal_output_verbose(5, orte_sensor_base_framework.framework_output,
+                                        "sensor:base:select Replacing component %s with %s - both measure %s",
+                                        tmp_module->component->base_version.mca_component_name,
+                                        component->base_version.mca_component_name,
+                                        component->data_measured);
+                    OBJ_RELEASE(tmp_module);
+                    opal_pointer_array_set_item(&tmp_array, i, NULL);
+                    break;
+                } else {
+                    duplicate = true;
+                }
+            }
+        }
+        if (duplicate) {
+            /* ignore this component */
+            opal_output_verbose(5, orte_sensor_base_framework.framework_output,
+                                "sensor:base:select Ignoring component %s - duplicate with higher priority measures %s",
+                                component->base_version.mca_component_name,
+                                component->data_measured);
             continue;
         }
 
@@ -102,7 +136,7 @@ int orte_sensor_base_select(void)
          */
         opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                             "sensor:base:select Query of component [%s] set priority to %d", 
-                            component->mca_component_name, priority);
+                            component->base_version.mca_component_name, priority);
         tmp_module = OBJ_NEW(orte_sensor_active_module_t);
         tmp_module->component = component;
         tmp_module->module    = (orte_sensor_base_module_t*)module;
@@ -169,7 +203,7 @@ int orte_sensor_base_select(void)
         }
         opal_output_verbose(5, orte_sensor_base_framework.framework_output,
                             "sensor:base:select Add module with priority [%s] %d", 
-                            tmp_module->component->mca_component_name, tmp_module->priority);
+                            tmp_module->component->base_version.mca_component_name, tmp_module->priority);
         opal_pointer_array_add(&orte_sensor_base.modules, tmp_module);
     }
     OBJ_DESTRUCT(&tmp_array);

orte/mca/sensor/base/sensor_private.h

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
  * Copyright (c) 2012      Los Alamos National Security, Inc. All rights reserved.
+ * Copyright (c) 2014      Intel, Inc. All rights reserved.
  *
  * $COPYRIGHT$
  *
@@ -50,7 +51,7 @@ typedef struct {
 
 typedef struct {
     opal_object_t super;
-    mca_base_component_t *component;
+    orte_sensor_base_component_t *component;
     orte_sensor_base_module_t *module;
     int priority;
 } orte_sensor_active_module_t;

orte/mca/sensor/coretemp/configure.m4

@@ -20,25 +20,20 @@ AC_DEFUN([MCA_orte_sensor_coretemp_CONFIG], [
 
     # do not build if support not requested
     AS_IF([test "$with_coretemp" != "no"],
-        [case "${host}" in
+          [AS_IF([test "$opal_found_linux" = "yes"],
-            i?86-*linux*|x86_64*linux*|ia64-*linux*|powerpc-*linux*|powerpc64-*linux*|sparc*-*linux*)
+                 [AS_IF([test -r "/sys/bus/platform/devices/coretemp.0"],
-            AS_IF([test -r "/sys/bus/platform/devices/coretemp.0"],
+                        [sensor_coretemp_happy=yes],
-                  [sensor_coretemp_happy=yes],
+                        [AC_MSG_WARN([Core temperature sensing was requested but the required directory])
-                  [AC_MSG_WARN([Core temperature sensing was requested but the required directory])
+                         AC_MSG_WARN([was not found. This usually indicates that the \"coretemp\"])
-                   AC_MSG_WARN([was not found. This usually indicates that the \"coretemp\"])
+                         AC_MSG_WARN([kernel module is not installed. Please install the module])
-                   AC_MSG_WARN([kernel module is not installed. Please install the module])
+                         AC_MSG_WARN([and try again, or remove the core temperature sensing request.])
-                   AC_MSG_WARN([and try again, or remove the core temperature sensing request.])
+                         sensor_coretemp_happy=no])],
-                   sensor_coretemp_happy=no])
+                 [AC_MSG_WARN([Core temperature sensing was requested but is only supported on Linux systems])
-            ;;
+                  sensor_coretemp_happy=no])
-            *)
+           AS_IF([test "$sensor_coretemp_happy" = "yes"],
-            AC_MSG_WARN([Core temperature sensing was requested but is only supported on Linux systems])
+                 [$1],
-            sensor_coretemp_happy=no
+                 [AC_MSG_ERROR([Cannot continue])
-            ;;
+                  $2])
-         esac
+          ],
-        AS_IF([test "$sensor_coretemp_happy" = "yes"],
+          [$2])
-            [$1],
-            [AC_MSG_ERROR([Cannot continue])
-             $2])
-        ],
-        [$2])
 ])dnl

orte/mca/sensor/coretemp/sensor_coretemp_component.c

@@ -40,7 +40,8 @@ orte_sensor_coretemp_component_t mca_sensor_coretemp_component = {
         {
             /* The component is checkpoint ready */
             MCA_BASE_METADATA_PARAM_CHECKPOINT
-        }
+        },
+        "coretemp"  // data being sensed
     }
 };
 

orte/mca/sensor/file/sensor_file_component.c

@@ -44,7 +44,8 @@ orte_sensor_file_component_t mca_sensor_file_component = {
         {
             /* The component is checkpoint ready */
             MCA_BASE_METADATA_PARAM_CHECKPOINT
-        }
+        },
+        "filemods"  // data being sensed
     }
 };
 

orte/mca/sensor/ft_tester/sensor_ft_tester_component.c

@@ -45,7 +45,8 @@ orte_sensor_ft_tester_component_t mca_sensor_ft_tester_component = {
         {
             /* The component is checkpoint ready */
             MCA_BASE_METADATA_PARAM_CHECKPOINT
-        }
+        },
+        NULL
     }
 };
 

orte/mca/sensor/heartbeat/sensor_heartbeat_component.c

@@ -43,7 +43,8 @@ orte_sensor_base_component_t mca_sensor_heartbeat_component = {
     {
         /* The component is checkpoint ready */
         MCA_BASE_METADATA_PARAM_CHECKPOINT
-    }
+    },
+    "heartbeat"
 };
 
 

orte/mca/sensor/resusage/sensor_resusage_component.c

@@ -44,7 +44,8 @@ orte_sensor_resusage_component_t mca_sensor_resusage_component = {
         {
             /* The component is checkpoint ready */
             MCA_BASE_METADATA_PARAM_CHECKPOINT
-        }
+        },
+        "procresource,noderesource"
     }
 };
 

orte/mca/sensor/sensor.h

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2009      Cisco Systems, Inc.  All rights reserved. 
  * Copyright (c) 2012      Los Alamos National Security, Inc. All rights reserved.
+ * Copyright (c) 2014      Intel, Inc. All rights reserved.
  *
  * $COPYRIGHT$
  *
@@ -81,6 +82,7 @@ typedef orte_sensor_base_module_1_0_0_t orte_sensor_base_module_t;
 struct orte_sensor_base_component_1_0_0_t {
     mca_base_component_t base_version;
     mca_base_component_data_t base_data;
+    char *data_measured;
 };
 typedef struct orte_sensor_base_component_1_0_0_t orte_sensor_base_component_1_0_0_t;
 typedef orte_sensor_base_component_1_0_0_t orte_sensor_base_component_t;

orte/mca/sensor/sigar/configure.m4

@@ -20,38 +20,32 @@ AC_DEFUN([MCA_orte_sensor_sigar_CONFIG], [
 
     # do not build if support not requested
     AS_IF([test "$with_sigar" != "no"],
-        [case "${host}" in
+          [AS_IF([test "$opal_found_linux" = "yes" || test "$opal_found_apple" = "yes"],
-            i?86-*linux*|x86_64*linux*|ia64-*linux*|powerpc-*linux*|powerpc64-*linux*|sparc*-*linux*)
+                 [AS_IF([test "$opal_found_apple" = "yes"], 
-            AS_IF([test -r "/proc/cpuinfo"],
+                        [libname="sigar-universal-macosx"], [libname="sigar"])
-                [sensor_linux_happy="yes"],
-                [sensor_linux_happy="no"])
-            ;;
-            *)
-            sensor_linux_happy="no"
-            ;;
-         esac
 
-         AS_IF([test "$sensor_linux_happy" = "yes"], 
+                  AS_IF([test ! -z "$with_sigar" -a "$with_sigar" != "yes"],
-               [libname="sigar"], [libname="sigar-universal-macosx"])
+                        [orte_check_sigar_dir="$with_sigar"])
 
-         AS_IF([test ! -z "$with_sigar" -a "$with_sigar" != "yes"],
+                  OMPI_CHECK_PACKAGE([sensor_sigar],
-               [orte_check_sigar_dir="$with_sigar"])
+                                     [sigar.h],
-
+                                     [$libname],
-         OMPI_CHECK_PACKAGE([sensor_sigar],
+                                     [sigar_proc_cpu_get],
-                            [sigar.h],
+                                     [],
-                            [$libname],
+                                     [$orte_check_sigar_dir],
-                            [sigar_proc_cpu_get],
+                                     [],
-                            [],
+                                     [$1],
-                            [$orte_check_sigar_dir],
+                                     [AC_MSG_WARN([SIGAR SENSOR SUPPORT REQUESTED])
-                            [],
+                                      AC_MSG_WARN([BUT REQUIRED LIBRARY OR HEADER NOT FOUND])
-                            [$1],
+                                      AC_MSG_ERROR([CANNOT CONTINUE])
-                            [AC_MSG_WARN([SIGAR SENSOR SUPPORT REQUESTED])
+                                      $2])],
-                             AC_MSG_WARN([BUT REQUIRED LIBRARY OR HEADER NOT FOUND])
+                 [AC_MSG_WARN([SIGAR SENSOR SUPPORT REQUESTED])
-                             AC_MSG_ERROR([CANNOT CONTINUE])
+                  AC_MSG_WARN([BUT ONLY SUPPORTED ON LINUX AND MAC])
-                             $2])],
+                  AC_MSG_ERROR([CANNOT CONTINUE])
+                  $2])],
           [$2])
 
-    AC_DEFINE_UNQUOTED(ORTE_SIGAR_LINUX, [test "$sensor_linux_happy" = "yes"],
+    AC_DEFINE_UNQUOTED(ORTE_SIGAR_LINUX, [test "$opal_found_linux" = "yes"],
                        [Which name to use for the sigar library on this OS])
     AC_SUBST(sensor_sigar_CPPFLAGS)
     AC_SUBST(sensor_sigar_LDFLAGS)

orte/mca/sensor/sigar/sensor_sigar_component.c

@@ -40,7 +40,8 @@ orte_sensor_sigar_component_t mca_sensor_sigar_component = {
         {
             /* The component is checkpoint ready */
             MCA_BASE_METADATA_PARAM_CHECKPOINT
-        }
+        },
+        "procresource,noderesource"
     }
 };
 
@@ -58,7 +59,7 @@ static int orte_sensor_sigar_query(mca_base_module_t **module, int *priority)
      * even if we aren't going to sample as we have to be
      * present in order to log any received results
      */
-    *priority = 50;  /* ahead of heartbeat */
+    *priority = 150;  /* ahead of heartbeat and resusage */
     *module = (mca_base_module_t *)&orte_sensor_sigar_module;
     return ORTE_SUCCESS;
 }