int
main(int argc, char *argv[])
{
int ret, exit_status = OPAL_SUCCESS;
int child_pid;
int prev_pid = 0;
int idx;
opal_crs_base_snapshot_t *snapshot = NULL;
char * tmp_env_var = NULL;
bool select = false;
/***************
* Initialize
***************/
if (OPAL_SUCCESS != (ret = initialize(argc, argv))) {
exit_status = ret;
goto cleanup;
}
/*
* Check for existence of the file, or program in the case of self
*/
if( OPAL_SUCCESS != (ret = check_file() )) {
opal_show_help("help-opal-restart.txt", "invalid_filename", true,
opal_restart_globals.snapshot_ref);
exit_status = ret;
goto cleanup;
}
/* Re-enable the selection of the CRS component, so we can choose the right one */
idx = mca_base_var_find(NULL, "crs", "base", "do_not_select");
if (0 > idx) {
opal_output(opal_restart_globals.output,
"MCA variable opal_crs_base_do_not_select not found\n");
exit_status = OPAL_ERROR;
goto cleanup;
}
ret = mca_base_var_set_value(idx, &select, 0, MCA_BASE_VAR_SOURCE_DEFAULT, NULL);
if (OPAL_SUCCESS != ret) {
exit_status = ret;
goto cleanup;
}
/*
* Make sure we are using the correct checkpointer
*/
if(NULL == expected_crs_comp) {
char * full_metadata_path = NULL;
FILE * metadata = NULL;
opal_asprintf(&full_metadata_path, "%s/%s/%s",
opal_restart_globals.snapshot_loc,
opal_restart_globals.snapshot_ref,
opal_restart_globals.snapshot_metadata);
if( NULL == (metadata = fopen(full_metadata_path, "r")) ) {
opal_show_help("help-opal-restart.txt", "invalid_metadata", true,
opal_restart_globals.snapshot_metadata,
full_metadata_path);
exit_status = OPAL_ERROR;
goto cleanup;
}
if( OPAL_SUCCESS != (ret = opal_crs_base_extract_expected_component(metadata,
&expected_crs_comp,
&prev_pid)) ) {
opal_show_help("help-opal-restart.txt", "invalid_metadata", true,
opal_restart_globals.snapshot_metadata,
full_metadata_path);
exit_status = ret;
goto cleanup;
}
free(full_metadata_path);
full_metadata_path = NULL;
fclose(metadata);
metadata = NULL;
}
opal_output_verbose(10, opal_restart_globals.output,
"Restart Expects checkpointer: (%s)",
expected_crs_comp);
(void) mca_base_var_env_name("crs", &tmp_env_var);
opal_setenv(tmp_env_var,
expected_crs_comp,
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/* Select this component or don't continue.
* If the selection of this component fails, then we can't
* restart on this node because it doesn't have the proper checkpointer
* available.
*/
if( OPAL_SUCCESS != (ret = opal_crs_base_open(MCA_BASE_OPEN_DEFAULT)) ) {
opal_show_help("help-opal-restart.txt", "comp_select_failure", true,
"crs", ret);
exit_status = ret;
goto cleanup;
}
if( OPAL_SUCCESS != (ret = opal_crs_base_select()) ) {
opal_show_help("help-opal-restart.txt", "comp_select_failure", true,
expected_crs_comp, ret);
exit_status = ret;
goto cleanup;
}
/*
* Make sure we have selected the proper component
*/
if(NULL == expected_crs_comp ||
0 != strncmp(expected_crs_comp,
opal_crs_base_selected_component.base_version.mca_component_name,
strlen(expected_crs_comp)) ) {
opal_show_help("help-opal-restart.txt", "comp_select_mismatch",
true,
expected_crs_comp,
opal_crs_base_selected_component.base_version.mca_component_name,
ret);
exit_status = ret;
goto cleanup;
}
/******************************
* Restart in this process
******************************/
opal_output_verbose(10, opal_restart_globals.output,
"Restarting from file (%s)\n",
opal_restart_globals.snapshot_ref);
snapshot = OBJ_NEW(opal_crs_base_snapshot_t);
snapshot->cold_start = true;
opal_asprintf(&(snapshot->snapshot_directory), "%s/%s",
opal_restart_globals.snapshot_loc,
opal_restart_globals.snapshot_ref);
opal_asprintf(&(snapshot->metadata_filename), "%s/%s",
snapshot->snapshot_directory,
opal_restart_globals.snapshot_metadata);
/* Since some checkpoint/restart systems don't pass along env vars to the
* restarted app, we need to take care of that.
*
* Included here is the creation of any files or directories that need to be
* created before the process is restarted.
*/
if(OPAL_SUCCESS != (ret = post_env_vars(prev_pid, snapshot) ) ) {
exit_status = ret;
goto cleanup;
}
/*
* Do the actual restart
*/
ret = opal_crs.crs_restart(snapshot,
false,
&child_pid);
if (OPAL_SUCCESS != ret) {
opal_show_help("help-opal-restart.txt", "restart_cmd_failure", true,
opal_restart_globals.snapshot_ref,
ret,
opal_crs_base_selected_component.base_version.mca_component_name);
exit_status = ret;
goto cleanup;
}
/* Should never get here, since crs_restart calls exec */
/***************
* Cleanup
***************/
cleanup:
if (OPAL_SUCCESS != (ret = finalize())) {
return ret;
}
if(NULL != snapshot )
OBJ_DESTRUCT(snapshot);
return exit_status;
}
/*
* Open all MCA components so that they can register their MCA
* parameters. Take a shotgun approach here and indiscriminately open
* all components -- don't be selective. To this end, we need to clear
* out the environment of all OMPI_MCA_<type> variables to ensure
* that the open algorithms don't try to only open one component.
*/
void orte_info_open_components(void)
{
int i;
char *env, *str;
char *target, *save, *type;
char **env_save=NULL;
bool need_close_components = false;
orte_info_component_map_t *map;
if (opened_components) {
return;
}
/* init the map */
OBJ_CONSTRUCT(&component_map, opal_pointer_array_t);
opal_pointer_array_init(&component_map, 256, INT_MAX, 128);
/* Clear out the environment. Use strdup() to orphan the resulting
* strings because items are placed in the environment by reference,
* not by value.
*/
for (i = 0; i < mca_types.size; ++i) {
if (NULL == (type = (char*)opal_pointer_array_get_item(&mca_types, i))) {
continue;
}
asprintf(&env, "OMPI_MCA_%s", type);
if (NULL != (save = getenv(env))) {
/* save this param so it can later be restored */
asprintf(&str, "%s=%s", env, save);
opal_argv_append_nosize(&env_save, str);
free(str);
/* can't manipulate it directly, so make a copy first */
asprintf(&target, "%s=", env);
putenv(target);
free(target);
}
}
/* some components require the event library be active, so activate it */
if (OPAL_SUCCESS != opal_event_base_open()) {
str = "opal_event_base_open";
goto error;
}
/* Open the DSS */
if (ORTE_SUCCESS != opal_dss_open()) {
str = "Unable to initialize the DSS";
goto error;
}
/* Open up the MCA */
if (OPAL_SUCCESS != mca_base_open()) {
str = "mca_base_open failed";
goto error;
}
/* Register the OPAL layer's MCA parameters */
if (OPAL_SUCCESS != opal_register_params()) {
str = "opal_register_params failed";
goto error;
}
/* Register the ORTE layer's MCA parameters */
if (ORTE_SUCCESS != orte_register_params()) {
str = "orte_register_params failed";
goto error;
}
/* Initialize the opal_output system */
if (!opal_output_init()) {
str = "opal_output_init failed";
goto error;
}
/* Find / open all components */
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("base");
opal_pointer_array_add(&component_map, map);
/* set default error message from here forward */
str = "A component framework failed to open properly.";
/* OPAL frameworks */
if (OPAL_SUCCESS != opal_backtrace_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("backtrace");
map->components = &opal_backtrace_base_components_opened;
opal_pointer_array_add(&component_map, map);
if (OPAL_SUCCESS != opal_memory_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("memory");
map->components = &opal_memory_base_components_opened;
opal_pointer_array_add(&component_map, map);
/* the event framework is already open - just get its components */
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("event");
map->components = &opal_event_components;
opal_pointer_array_add(&component_map, map);
if (OPAL_SUCCESS != opal_memchecker_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("memchecker");
map->components = &opal_memchecker_base_components_opened;
opal_pointer_array_add(&component_map, map);
if (OPAL_SUCCESS != opal_shmem_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("shmem");
map->components = &opal_shmem_base_components_opened;
opal_pointer_array_add(&component_map, map);
#if OPAL_HAVE_HWLOC
if (OPAL_SUCCESS != opal_hwloc_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("hwloc");
map->components = &opal_hwloc_base_components;
opal_pointer_array_add(&component_map, map);
#endif
if (OPAL_SUCCESS != opal_timer_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("timer");
map->components = &opal_timer_base_components_opened;
opal_pointer_array_add(&component_map, map);
#if OPAL_ENABLE_FT_CR == 1
if (OPAL_SUCCESS != opal_crs_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("crs");
map->components = &opal_crs_base_components_available;
opal_pointer_array_add(&component_map, map);
#endif
if (OPAL_SUCCESS != opal_if_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("if");
map->components = &opal_if_components;
opal_pointer_array_add(&component_map, map);
/* OPAL's installdirs base open has already been called as part of
* opal_init_util() back in main().
*/
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("installdirs");
map->components = &opal_installdirs_components;
opal_pointer_array_add(&component_map, map);
/* ORTE frameworks
* Set orte_process_info.proc_type to HNP to force all frameworks to
* open components
*/
orte_process_info.proc_type = ORTE_PROC_HNP;
if (ORTE_SUCCESS != orte_state_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("state");
map->components = &orte_state_base_components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_errmgr_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("errmgr");
map->components = &orte_errmgr_base_components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_grpcomm_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("grpcomm");
map->components = &orte_grpcomm_base.components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_db_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("db");
map->components = &orte_db_base.available_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_ess_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("ess");
map->components = &orte_ess_base_components_available;
opal_pointer_array_add(&component_map, map);
#if !ORTE_DISABLE_FULL_SUPPORT
if (ORTE_SUCCESS != mca_oob_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("oob");
map->components = &mca_oob_base_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_odls_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("odls");
map->components = &orte_odls_base.available_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_iof_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("iof");
map->components = &orte_iof_base.iof_components_opened;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_ras_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("ras");
map->components = &orte_ras_base.ras_opened;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_rmaps_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("rmaps");
map->components = &orte_rmaps_base.available_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_rml_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("rml");
map->components = &orte_rml_base_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_routed_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("routed");
map->components = &orte_routed_base_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_plm_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("plm");
map->components = &orte_plm_base.available_components;
opal_pointer_array_add(&component_map, map);
#if OPAL_ENABLE_FT_CR == 1
if (ORTE_SUCCESS != orte_snapc_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("snapc");
map->components = &orte_snapc_base_components_available;
opal_pointer_array_add(&component_map, map);
#endif
if (ORTE_SUCCESS != orte_sensor_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("sensor");
map->components = &mca_sensor_base_components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_filem_base_open()) {
goto error;
}
map = OBJ_NEW(orte_info_component_map_t);
map->type = strdup("filem");
map->components = &orte_filem_base_components_available;
opal_pointer_array_add(&component_map, map);
#endif
/* flag that we need to close components */
need_close_components = true;
/* Restore the environment to what it was before we started so that
* if users setenv OMPI_MCA_<framework name> to some value, they'll
* see that value when it is shown via --param output.
*/
if (NULL != env_save) {
for (i = 0; i < opal_argv_count(env_save); ++i) {
putenv(env_save[i]);
}
}
/* All done */
opened_components = true;
return;
error:
fprintf(stderr, "%s\n", str);
fprintf(stderr, "orte-info will likely not display all configuration information\n");
if (need_close_components) {
opened_components = true;
orte_info_close_components();
}
}
int opal_cr_init(void )
{
int ret, exit_status = OPAL_SUCCESS;
opal_cr_coord_callback_fn_t prev_coord_func;
int val;
if( ++opal_cr_initalized != 1 ) {
if( opal_cr_initalized < 1 ) {
exit_status = OPAL_ERROR;
goto cleanup;
}
exit_status = OPAL_SUCCESS;
goto cleanup;
}
/*
* Some startup MCA parameters
*/
ret = mca_base_param_reg_int_name("opal_cr", "verbose",
"Verbose output level for the runtime OPAL Checkpoint/Restart functionality",
false, false,
0,
&val);
if(0 != val) {
opal_cr_output = opal_output_open(NULL);
} else {
opal_cr_output = -1;
}
opal_output_set_verbosity(opal_cr_output, val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Verbose Level: %d",
val);
mca_base_param_reg_int_name("ft", "cr_enabled",
"Enable fault tolerance for this program",
false, false,
0, &val);
opal_cr_set_enabled(OPAL_INT_TO_BOOL(val));
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT Enabled: %d",
val);
mca_base_param_reg_int_name("opal_cr", "enable_timer",
"Enable Checkpoint timer (Default: Disabled)",
false, false,
0, &val);
opal_cr_timing_enabled = OPAL_INT_TO_BOOL(val);
mca_base_param_reg_int_name("opal_cr", "enable_timer_barrier",
"Enable Checkpoint timer Barrier (Default: Disabled)",
false, false,
0, &val);
if( opal_cr_timing_enabled ) {
opal_cr_timing_barrier_enabled = OPAL_INT_TO_BOOL(val);
} else {
opal_cr_timing_barrier_enabled = false;
}
mca_base_param_reg_int_name("opal_cr", "timer_target_rank",
"Target Rank for the timer (Default: 0)",
false, false,
0, &val);
opal_cr_timing_target_rank = val;
#if OPAL_ENABLE_FT_THREAD == 1
mca_base_param_reg_int_name("opal_cr", "use_thread",
"Use an async thread to checkpoint this program (Default: Disabled)",
false, false,
0, &val);
opal_cr_thread_use_if_avail = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT Use thread: %d",
val);
mca_base_param_reg_int_name("opal_cr", "thread_sleep_check",
"Time to sleep between checking for a checkpoint (Default: 0)",
false, false,
0, &val);
opal_cr_thread_sleep_check = val;
mca_base_param_reg_int_name("opal_cr", "thread_sleep_wait",
"Time to sleep waiting for process to exit MPI library (Default: 0)",
false, false,
0, &val);
opal_cr_thread_sleep_wait = val;
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT thread sleep: check = %d, wait = %d",
opal_cr_thread_sleep_check, opal_cr_thread_sleep_wait);
#endif
mca_base_param_reg_int_name("opal_cr", "is_tool",
"Is this a tool program, meaning does it require a fully operational OPAL or just enough to exec.",
false, false,
0,
&val);
opal_cr_is_tool = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Is a tool program: %d",
val);
#ifndef __WINDOWS__
mca_base_param_reg_int_name("opal_cr", "signal",
"Checkpoint/Restart signal used to initialize an OPAL Only checkpoint of a program",
false, false,
SIGUSR1,
&opal_cr_entry_point_signal);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Checkpoint Signal: %d",
opal_cr_entry_point_signal);
mca_base_param_reg_int_name("opal_cr", "debug_sigpipe",
"Activate a signal handler for debugging SIGPIPE Errors that can happen on restart. (Default: Disabled)",
false, false,
0, &val);
opal_cr_debug_sigpipe = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Debug SIGPIPE: %d (%s)",
val, (opal_cr_debug_sigpipe ? "True" : "False"));
#if OPAL_ENABLE_FT_THREAD == 1
/* If we have a thread, then attach the SIGPIPE signal handler there since
* it is most likely to be the one that needs it.
*/
if( opal_cr_debug_sigpipe && !opal_cr_thread_use_if_avail ) {
if( SIG_ERR == signal(SIGPIPE, opal_cr_sigpipe_debug_signal_handler) ) {
;
}
}
#else
if( opal_cr_debug_sigpipe ) {
if( SIG_ERR == signal(SIGPIPE, opal_cr_sigpipe_debug_signal_handler) ) {
;
}
}
#endif
#else
opal_cr_is_tool = true; /* no support for CR on Windows yet */
#endif /* __WINDOWS__ */
mca_base_param_reg_string_name("opal_cr", "tmp_dir",
"Temporary directory to place rendezvous files for a checkpoint",
false, false,
"/tmp",
&opal_cr_pipe_dir);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Temp Directory: %s",
opal_cr_pipe_dir);
if( !opal_cr_is_tool ) {
/* Register the OPAL interlevel coordination callback */
opal_cr_reg_coord_callback(opal_cr_coord, &prev_coord_func);
opal_cr_stall_check = false;
opal_cr_currently_stalled = false;
} /* End opal_cr_is_tool = true */
/*
* If fault tolerance was not compiled in then
* we need to make sure that the listener thread is active to tell
* the tools that this is not a checkpointable job.
* We don't need the CRS framework to be initalized.
*/
#if OPAL_ENABLE_FT == 1
/*
* Open the checkpoint / restart service components
*/
if (OPAL_SUCCESS != (ret = opal_crs_base_open())) {
opal_output(opal_cr_output,
"opal_cr: init: opal_crs_base_open Failed to open. (%d)\n", ret);
exit_status = ret;
goto cleanup;
}
if (OPAL_SUCCESS != (ret = opal_crs_base_select())) {
opal_output(opal_cr_output,
"opal_cr: init: opal_crs_base_select Failed. (%d)\n", ret);
exit_status = ret;
goto cleanup;
}
#endif
#if OPAL_ENABLE_FT_THREAD == 1
if( !opal_cr_is_tool && opal_cr_thread_use_if_avail) {
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: starting the thread\n");
opal_set_using_threads(true);
/*
* Start the thread
*/
OBJ_CONSTRUCT(&opal_cr_thread, opal_thread_t);
OBJ_CONSTRUCT(&opal_cr_thread_lock, opal_mutex_t);
opal_cr_thread_is_done = false;
opal_cr_thread_is_active = false;
opal_cr_thread_in_library = false;
opal_cr_thread_num_in_library = 0;
opal_cr_thread.t_run = opal_cr_thread_fn;
opal_cr_thread.t_arg = NULL;
opal_thread_start(&opal_cr_thread);
} /* End opal_cr_is_tool = true */
else {
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: *Not* Using C/R thread\n");
}
#endif /* OPAL_ENABLE_FT_THREAD == 1 */
cleanup:
return exit_status;
}
int opal_cr_init(void )
{
int ret, exit_status = OPAL_SUCCESS;
opal_cr_coord_callback_fn_t prev_coord_func;
int val, t;
if( ++opal_cr_initalized != 1 ) {
if( opal_cr_initalized < 1 ) {
exit_status = OPAL_ERROR;
goto cleanup;
}
exit_status = OPAL_SUCCESS;
goto cleanup;
}
/*
* Some startup MCA parameters
*/
ret = mca_base_param_reg_int_name("opal_cr", "verbose",
"Verbose output level for the runtime OPAL Checkpoint/Restart functionality",
false, false,
0,
&val);
if(0 != val) {
opal_cr_output = opal_output_open(NULL);
} else {
opal_cr_output = -1;
}
opal_output_set_verbosity(opal_cr_output, val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Verbose Level: %d",
val);
mca_base_param_reg_int_name("ft", "cr_enabled",
"Enable fault tolerance for this program",
false, false,
0, &val);
opal_cr_set_enabled(OPAL_INT_TO_BOOL(val));
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT Enabled: %d",
val);
mca_base_param_reg_int_name("opal_cr", "enable_timer",
"Enable Checkpoint timer (Default: Disabled)",
false, false,
0, &val);
opal_cr_timing_enabled = OPAL_INT_TO_BOOL(val);
mca_base_param_reg_int_name("opal_cr", "enable_timer_barrier",
"Enable Checkpoint timer Barrier (Default: Disabled)",
false, false,
0, &val);
if( opal_cr_timing_enabled ) {
opal_cr_timing_barrier_enabled = OPAL_INT_TO_BOOL(val);
} else {
opal_cr_timing_barrier_enabled = false;
}
mca_base_param_reg_int_name("opal_cr", "timer_target_rank",
"Target Rank for the timer (Default: 0)",
false, false,
0, &val);
opal_cr_timing_target_rank = val;
#if OPAL_ENABLE_FT_THREAD == 1
mca_base_param_reg_int_name("opal_cr", "use_thread",
"Use an async thread to checkpoint this program (Default: Disabled)",
false, false,
0, &val);
opal_cr_thread_use_if_avail = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT Use thread: %d",
val);
mca_base_param_reg_int_name("opal_cr", "thread_sleep_check",
"Time to sleep between checking for a checkpoint (Default: 0)",
false, false,
0, &val);
opal_cr_thread_sleep_check = val;
mca_base_param_reg_int_name("opal_cr", "thread_sleep_wait",
"Time to sleep waiting for process to exit MPI library (Default: 1000)",
false, false,
1000, &val);
opal_cr_thread_sleep_wait = val;
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: FT thread sleep: check = %d, wait = %d",
opal_cr_thread_sleep_check, opal_cr_thread_sleep_wait);
#endif
mca_base_param_reg_int_name("opal_cr", "is_tool",
"Is this a tool program, meaning does it require a fully operational OPAL or just enough to exec.",
false, false,
0,
&val);
opal_cr_is_tool = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Is a tool program: %d",
val);
#if OPAL_ENABLE_CRDEBUG == 1
mca_base_param_reg_int_name("opal_cr", "enable_crdebug",
"Enable checkpoint/restart debugging",
false, false,
0,
&val);
MPIR_debug_with_checkpoint = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: C/R Debugging Enabled [%s]\n",
(MPIR_debug_with_checkpoint ? "True": "False"));
#endif
#ifndef __WINDOWS__
mca_base_param_reg_int_name("opal_cr", "signal",
"Checkpoint/Restart signal used to initialize an OPAL Only checkpoint of a program",
false, false,
SIGUSR1,
&opal_cr_entry_point_signal);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Checkpoint Signal: %d",
opal_cr_entry_point_signal);
mca_base_param_reg_int_name("opal_cr", "debug_sigpipe",
"Activate a signal handler for debugging SIGPIPE Errors that can happen on restart. (Default: Disabled)",
false, false,
0, &val);
opal_cr_debug_sigpipe = OPAL_INT_TO_BOOL(val);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Debug SIGPIPE: %d (%s)",
val, (opal_cr_debug_sigpipe ? "True" : "False"));
#if OPAL_ENABLE_FT_THREAD == 1
/* If we have a thread, then attach the SIGPIPE signal handler there since
* it is most likely to be the one that needs it.
*/
if( opal_cr_debug_sigpipe && !opal_cr_thread_use_if_avail ) {
if( SIG_ERR == signal(SIGPIPE, opal_cr_sigpipe_debug_signal_handler) ) {
;
}
}
#else
if( opal_cr_debug_sigpipe ) {
if( SIG_ERR == signal(SIGPIPE, opal_cr_sigpipe_debug_signal_handler) ) {
;
}
}
#endif
#else
opal_cr_is_tool = true; /* no support for CR on Windows yet */
#endif /* __WINDOWS__ */
#if OPAL_ENABLE_CRDEBUG == 1
opal_cr_debug_num_free_threads = 3;
opal_cr_debug_free_threads = (opal_thread_t **)malloc(sizeof(opal_thread_t *) * opal_cr_debug_num_free_threads );
for(t = 0; t < opal_cr_debug_num_free_threads; ++t ) {
opal_cr_debug_free_threads[t] = NULL;
}
mca_base_param_reg_int_name("opal_cr", "crdebug_signal",
"Checkpoint/Restart signal used to hold threads when debugging",
false, false,
SIGTSTP,
&opal_cr_debug_signal);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Checkpoint Signal (Debug): %d",
opal_cr_debug_signal);
if( SIG_ERR == signal(opal_cr_debug_signal, MPIR_checkpoint_debugger_signal_handler) ) {
opal_output(opal_cr_output,
"opal_cr: init: Failed to register C/R debug signal (%d)",
opal_cr_debug_signal);
}
#else
/* Silence a compiler warning */
t = 0;
#endif
mca_base_param_reg_string_name("opal_cr", "tmp_dir",
"Temporary directory to place rendezvous files for a checkpoint",
false, false,
opal_tmp_directory(),
&opal_cr_pipe_dir);
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: Temp Directory: %s",
opal_cr_pipe_dir);
if( !opal_cr_is_tool ) {
/* Register the OPAL interlevel coordination callback */
opal_cr_reg_coord_callback(opal_cr_coord, &prev_coord_func);
opal_cr_stall_check = false;
opal_cr_currently_stalled = false;
} /* End opal_cr_is_tool = true */
/*
* If fault tolerance was not compiled in then
* we need to make sure that the listener thread is active to tell
* the tools that this is not a checkpointable job.
* We don't need the CRS framework to be initalized.
*/
#if OPAL_ENABLE_FT_CR == 1
/*
* Open the checkpoint / restart service components
*/
if (OPAL_SUCCESS != (ret = opal_crs_base_open())) {
opal_show_help( "help-opal-runtime.txt",
"opal_cr_init:no-crs", true,
"opal_crs_base_open", ret );
exit_status = ret;
goto cleanup;
}
if (OPAL_SUCCESS != (ret = opal_crs_base_select())) {
opal_show_help( "help-opal-runtime.txt",
"opal_cr_init:no-crs", true,
"opal_crs_base_select", ret );
exit_status = ret;
goto cleanup;
}
#endif
#if OPAL_ENABLE_FT_THREAD == 1
if( !opal_cr_is_tool && opal_cr_thread_use_if_avail) {
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: starting the thread\n");
/* JJH: We really do need this line below since it enables
* actual locks for threads. However currently the
* upper layers will deadlock if it is enabled.
* So hack around the problem for now, while working
* on a complete solution. See ticket #2741 for more
* details.
* opal_set_using_threads(true);
*/
/*
* Start the thread
*/
OBJ_CONSTRUCT(&opal_cr_thread, opal_thread_t);
OBJ_CONSTRUCT(&opal_cr_thread_lock, opal_mutex_t);
opal_cr_thread_is_done = false;
opal_cr_thread_is_active = false;
opal_cr_thread_in_library = false;
opal_cr_thread_num_in_library = 0;
opal_cr_thread.t_run = opal_cr_thread_fn;
opal_cr_thread.t_arg = NULL;
opal_thread_start(&opal_cr_thread);
} /* End opal_cr_is_tool = true */
else {
opal_output_verbose(10, opal_cr_output,
"opal_cr: init: *Not* Using C/R thread\n");
}
#endif /* OPAL_ENABLE_FT_THREAD == 1 */
cleanup:
return exit_status;
}
