Fix the newly added FT event state when compiling --with-ft

This commit was SVN r30988.

orte/mca/oob/base/oob_base_frame.c

@@ -30,6 +30,10 @@
 #include "orte/mca/rml/base/base.h"
 #include "orte/mca/oob/base/base.h"
 
+#if OPAL_ENABLE_FT_CR == 1
+#include "orte/mca/state/state.h"
+#endif
+
 /*
  * The following file was created by configure.  It contains extern
  * statements and the definition of an array of pointers to each
@@ -88,8 +92,10 @@ static int orte_oob_base_open(mca_base_open_flag_t flags)
     OBJ_CONSTRUCT(&orte_oob_base.actives, opal_list_t);
 
 #if OPAL_ENABLE_FT_CR == 1
-    /* register the FT event callback */
+    /* register the FT events callback */
-    orte_state.add_job_state(ORTE_JOB_STATE_FT_EVENT, orte_oob_ft_event, ORTE_ERROR_PRI);
+    orte_state.add_job_state(ORTE_JOB_STATE_FT_CHECKPOINT, orte_oob_base_ft_event, ORTE_ERROR_PRI);
+    orte_state.add_job_state(ORTE_JOB_STATE_FT_CONTINUE, orte_oob_base_ft_event, ORTE_ERROR_PRI);
+    orte_state.add_job_state(ORTE_JOB_STATE_FT_RESTART, orte_oob_base_ft_event, ORTE_ERROR_PRI);
 #endif
 
      /* Open up all available components */

orte/mca/oob/base/oob_base_stubs.c

@@ -383,17 +383,22 @@ static void process_uri(char *uri)
 void orte_oob_base_ft_event(int sd, short argc, void *cbdata)
 {
     int rc;
-    mca_oob_module_t *mod;
+    mca_base_component_list_item_t *cli;
+    mca_oob_base_component_t *component;
     orte_state_caddy_t *state = (orte_state_caddy_t*)cbdata;
 
     /* loop across all available modules in priority order
      * and call each one's ft_event handler
      */
-    OPAL_LIST_FOREACH(mod, &mca_oob_base_modules, mca_oob_module_t) {
+    OPAL_LIST_FOREACH(cli, &orte_oob_base.actives, mca_base_component_list_item_t) {
-        if (NULL != mod->ft_event) {
+        component = (mca_oob_base_component_t*)cli->cli_component;
-            if (ORTE_SUCCESS != (rc = mod->ft_event(state))) {
+        if (NULL == component->ft_event) {
-                ORTE_ERROR_LOG(rc);
+            /* doesn't support this ability */
-            }
+            continue;
+        }
+
+        if (ORTE_SUCCESS != (rc = component->ft_event(state->job_state))) {
+            ORTE_ERROR_LOG(rc);
         }
     }
     OBJ_RELEASE(state);

orte/mca/oob/oob.h

@@ -55,6 +55,9 @@ typedef int (*mca_oob_base_component_set_addr_fn_t)(orte_process_name_t *peer,
                                                     char **uris);
 typedef bool (*mca_oob_base_component_is_reachable_fn_t)(orte_process_name_t *peer);
 typedef void (*mca_oob_ping_callback_fn_t)(int status, void *cbdata);
+#if OPAL_ENABLE_FT_CR == 1
+typedef int (*mca_oob_base_component_ft_event_fn_t)(int state);
+#endif
 
 typedef struct {
     mca_base_component_t                      oob_base;
@@ -68,6 +71,9 @@ typedef struct {
     mca_oob_base_component_get_addr_fn_t      get_addr;
     mca_oob_base_component_set_addr_fn_t      set_addr;
     mca_oob_base_component_is_reachable_fn_t  is_reachable;
+#if OPAL_ENABLE_FT_CR == 1
+    mca_oob_base_component_ft_event_fn_t      ft_event;
+#endif
 } mca_oob_base_component_t;
 
 /**

orte/mca/plm/plm_types.h

@@ -174,7 +174,9 @@ typedef int32_t orte_job_state_t;
 #define ORTE_JOB_STATE_ALLOC_FAILED            (ORTE_JOB_STATE_ERROR + 18)  /* job failed to obtain an allocation */
 
 /* define an FT event */
-#define ORTE_JOB_STATE_FT_EVENT                (ORTE_JOB_STATE_ERROR + 19)
+#define ORTE_JOB_STATE_FT_CHECKPOINT           (ORTE_JOB_STATE_ERROR + 19)
+#define ORTE_JOB_STATE_FT_CONTINUE             (ORTE_JOB_STATE_ERROR + 20)
+#define ORTE_JOB_STATE_FT_RESTART              (ORTE_JOB_STATE_ERROR + 21)
 
 
 /* Define a boundary so that external developers