/**
* component open/close/init function
*/
static int orte_sensor_file_open(void)
{
mca_base_component_t *c = &mca_sensor_file_component.super.base_version;
int tmp;
/* lookup parameters */
mca_base_param_reg_string(c, "filename",
"File to be monitored",
false, false, NULL, &mca_sensor_file_component.file);
mca_base_param_reg_int(c, "check_size",
"Check the file size",
false, false, false, &tmp);
mca_sensor_file_component.check_size = OPAL_INT_TO_BOOL(tmp);
mca_base_param_reg_int(c, "check_access",
"Check access time",
false, false, false, &tmp);
mca_sensor_file_component.check_access = OPAL_INT_TO_BOOL(tmp);
mca_base_param_reg_int(c, "check_mod",
"Check modification time",
false, false, false, &tmp);
mca_sensor_file_component.check_mod = OPAL_INT_TO_BOOL(tmp);
mca_base_param_reg_int(c, "limit",
"Number of times the sensor can detect no motion before declaring error (default=3)",
false, false, 3, &mca_sensor_file_component.limit);
return ORTE_SUCCESS;
}
/*
* Register some paffinity-wide MCA params
*/
int opal_paffinity_base_register_params(void)
{
int value, id;
static int been_here = 0;
/* We may get called twice; be harmless in that case. */
if (1 == been_here) {
return OPAL_SUCCESS;
}
been_here = 1;
/* Debugging / verbose output */
mca_base_param_reg_int_name("paffinity", "base_verbose",
"Verbosity level of the paffinity framework",
false, false,
0, &value);
if (0 != value) {
opal_paffinity_base_output = opal_output_open(NULL);
} else {
opal_paffinity_base_output = -1;
}
id = mca_base_param_reg_int_name("opal", "paffinity_alone",
"If nonzero, assume that this job is the only (set of) process(es) running on each node and bind processes to processors, starting with processor ID 0",
false, false,
0, NULL);
/* register the historical mpi_paffinity_alone synonym, but don't
* declare it deprecated so we don't scare the users.
*
* Yes, this breaks the abstraction barrier, but as indicated
* on the developer list....live with it. :-)
*/
mca_base_param_reg_syn_name(id, "mpi", "paffinity_alone", false);
mca_base_param_lookup_int(id, &value);
opal_paffinity_alone = OPAL_INT_TO_BOOL(value);
mca_base_param_reg_int_name("paffinity", "base_bound",
"Process affinity was set by an external entity",
true, false,
false, &value);
opal_paffinity_base_bound = OPAL_INT_TO_BOOL(value);
mca_base_param_reg_string_name("paffinity", "base_applied_binding",
"Process affinity was set by an external entity",
true, false,
NULL, &opal_paffinity_base_applied_binding);
return OPAL_SUCCESS;
}
static int
mmap_register(void)
{
int value;
/*
* Do we want the "warning: your mmap file is on NFS!" message? Per a
* thread on the OMPI devel list
* (http://www.open-mpi.org/community/lists/devel/2011/12/10054.php),
* on some systems, it doesn't seem to matter. But per older threads,
* it definitely does matter on some systems. Perhaps newer kernels
* are smarter about this kind of stuff...? Regardless, we should
* provide the ability to turn off this message for systems where the
* effect doesn't matter.
*/
mca_base_param_reg_int(&mca_shmem_mmap_component.super.base_version,
"enable_nfs_warning",
"Enable the warning emitted when Open MPI detects "
"that its shared memory backing file is located on "
"a network filesystem (1 = enabled, 0 = disabled).",
false, false,
(int)true, &value);
opal_shmem_mmap_nfs_warning = OPAL_INT_TO_BOOL(value);
return OPAL_SUCCESS;
}
static int snapc_full_open(void)
{
int value;
/*
* This should be the last componet to ever get used since
* it doesn't do anything.
*/
mca_base_param_reg_int(&mca_snapc_full_component.super.base_version,
"priority",
"Priority of the SNAPC full component",
false, false,
mca_snapc_full_component.super.priority,
&mca_snapc_full_component.super.priority);
mca_base_param_reg_int(&mca_snapc_full_component.super.base_version,
"verbose",
"Verbose level for the SNAPC full component",
false, false,
mca_snapc_full_component.super.verbose,
&mca_snapc_full_component.super.verbose);
/* If there is a custom verbose level for this component than use it
* otherwise take our parents level and output channel
*/
if ( 0 != mca_snapc_full_component.super.verbose) {
mca_snapc_full_component.super.output_handle = opal_output_open(NULL);
opal_output_set_verbosity(mca_snapc_full_component.super.output_handle,
mca_snapc_full_component.super.verbose);
} else {
mca_snapc_full_component.super.output_handle = orte_snapc_base_output;
}
mca_base_param_reg_int(&mca_snapc_full_component.super.base_version,
"skip_filem",
"Not for general use! For debugging only! Pretend to move files. [Default = disabled]",
false, false,
0,
&value);
orte_snapc_full_skip_filem = OPAL_INT_TO_BOOL(value);
/*
* Debug Output
*/
opal_output_verbose(10, mca_snapc_full_component.super.output_handle,
"snapc:full: open()");
opal_output_verbose(20, mca_snapc_full_component.super.output_handle,
"snapc:full: open: priority = %d",
mca_snapc_full_component.super.priority);
opal_output_verbose(20, mca_snapc_full_component.super.output_handle,
"snapc:full: open: verbosity = %d",
mca_snapc_full_component.super.verbose);
opal_output_verbose(20, mca_snapc_full_component.super.output_handle,
"snapc:full: open: skip_filem = %s",
(orte_snapc_full_skip_filem == true ? "True" : "False"));
return ORTE_SUCCESS;
}
int opal_crs_base_select(void)
{
int ret, exit_status = OPAL_SUCCESS;
opal_crs_base_component_t *best_component = NULL;
opal_crs_base_module_t *best_module = NULL;
int int_value = 0;
/*
* Note: If we are a tool, then we will manually run the selection routine
* for the checkpointer. The tool will set the MCA parameter
* 'crs_base_do_not_select' before opal_init and then reset it after to
* disable the selection logic.
* This is useful for opal_restart because it reads the metadata file
* that indicates the checkpointer to be used after calling opal_init.
* Therefore it would need to select a specific module, but it doesn't
* know which one until later. It will set the MCA parameter 'crs'
* before calling this function.
*/
mca_base_param_reg_int_name("crs",
"base_do_not_select",
"Do not do the selection of the CRS component",
true, false,
false,
&int_value);
if( OPAL_INT_TO_BOOL(int_value) ) {
opal_output_verbose(10, opal_crs_base_output,
"crs:select: Not selecting at this time!");
return OPAL_SUCCESS;
}
/*
* Select the best component
*/
if( OPAL_SUCCESS != mca_base_select("crs", opal_crs_base_output,
&opal_crs_base_components_available,
(mca_base_module_t **) &best_module,
(mca_base_component_t **) &best_component) ) {
/* This will only happen if no component was selected */
exit_status = OPAL_ERROR;
goto cleanup;
}
/* Save the winner */
opal_crs_base_selected_component = *best_component;
opal_crs = *best_module;
/* Initialize the winner */
if (NULL != best_module) {
if (OPAL_SUCCESS != (ret = opal_crs.crs_init()) ) {
exit_status = ret;
goto cleanup;
}
}
cleanup:
return exit_status;
}
static int state_novm_open(void)
{
int tmp;
mca_base_component_t *c=&mca_state_novm_component.base_version;
mca_base_param_reg_int(c, "select",
"Use this component",
false, false, (int)false, &tmp);
select_me = OPAL_INT_TO_BOOL(tmp);
return ORTE_SUCCESS;
}
static int crs_blcr_open(void)
{
int value;
mca_base_param_reg_int(&mca_crs_blcr_component.super.base_version,
"priority",
"Priority of the CRS blcr component",
false, false,
mca_crs_blcr_component.super.priority,
&mca_crs_blcr_component.super.priority);
mca_base_param_reg_int(&mca_crs_blcr_component.super.base_version,
"verbose",
"Verbose level for the CRS blcr component",
false, false,
mca_crs_blcr_component.super.verbose,
&mca_crs_blcr_component.super.verbose);
/* If there is a custom verbose level for this component than use it
* otherwise take our parents level and output channel
*/
if ( 0 != mca_crs_blcr_component.super.verbose) {
mca_crs_blcr_component.super.output_handle = opal_output_open(NULL);
opal_output_set_verbosity(mca_crs_blcr_component.super.output_handle,
mca_crs_blcr_component.super.verbose);
} else {
mca_crs_blcr_component.super.output_handle = opal_crs_base_output;
}
mca_base_param_reg_int(&mca_crs_blcr_component.super.base_version,
"dev_null",
"Not for general use! For debugging only! Save checkpoint to /dev/null. [Default = disabled]",
false, false,
0,
&value);
opal_crs_blcr_dev_null = OPAL_INT_TO_BOOL(value);
/*
* Debug output
*/
opal_output_verbose(10, mca_crs_blcr_component.super.output_handle,
"crs:blcr: open()");
opal_output_verbose(20, mca_crs_blcr_component.super.output_handle,
"crs:blcr: open: priority = %d",
mca_crs_blcr_component.super.priority);
opal_output_verbose(20, mca_crs_blcr_component.super.output_handle,
"crs:blcr: open: verbosity = %d",
mca_crs_blcr_component.super.verbose);
opal_output_verbose(10, mca_crs_blcr_component.super.output_handle,
"crs:blcr: open: dev_null = %s",
(opal_crs_blcr_dev_null == true ? "True" : "False"));
return OPAL_SUCCESS;
}
/**
* Function for finding and opening either all MCA components, or the one
* that was specifically requested via a MCA parameter.
*/
int mca_btl_base_open(void)
{
int i;
if( ++mca_btl_base_already_opened > 1 ) return OMPI_SUCCESS;
/* Verbose output */
mca_base_param_reg_int_name("btl",
"base_verbose",
"Verbosity level of the BTL framework",
false, false,
0,
&mca_btl_base_verbose);
mca_btl_base_output = opal_output_open(NULL);
opal_output_set_verbosity(mca_btl_base_output, mca_btl_base_verbose);
/* Override the per-BTL "don't run if THREAD_MULTIPLE selected"
embargo? */
mca_base_param_reg_int_name("btl",
"base_thread_multiple_override",
"Enable BTLs that are not normally enabled when MPI_THREAD_MULTIPLE is enabled (THIS IS FOR DEVELOPERS ONLY! SHOULD NOT BE USED BY END USERS!)",
true, false,
0, &i);
mca_btl_base_thread_multiple_override = OPAL_INT_TO_BOOL(i);
/* Open up all available components */
if (OMPI_SUCCESS !=
mca_base_components_open("btl", mca_btl_base_output, mca_btl_base_static_components,
&mca_btl_base_components_opened, true)) {
return OMPI_ERROR;
}
/* Initialize the list so that in mca_btl_base_close(), we can
iterate over it (even if it's empty, as in the case of
ompi_info) */
OBJ_CONSTRUCT(&mca_btl_base_modules_initialized, opal_list_t);
/* register parameters */
mca_base_param_lookup_string(
mca_base_param_register_string("btl","base","include",NULL,NULL), &mca_btl_base_include);
mca_base_param_lookup_string(
mca_base_param_register_string("btl","base","exclude",NULL,NULL), &mca_btl_base_exclude);
mca_base_param_reg_int_name("btl", "base_warn_component_unused",
"This parameter is used to turn on warning messages when certain NICs are not used",
false, false, 1, &mca_btl_base_warn_component_unused);
/* All done */
return OMPI_SUCCESS;
}
/**
* Function for finding and opening either all MCA components, or the one
* that was specifically requested via a MCA parameter.
*/
int orte_debugger_base_open(void)
{
int value;
/* Debugging / verbose output. Always have stream open, with
verbose set by the mca open system... */
orte_debugger_base.output = opal_output_open(NULL);
mca_base_param_reg_int_name("orte",
"output_debugger_proctable",
"Whether or not to output the debugger proctable after launch (default: false)",
true, false, 0, &value);
orte_debugger_base.dump_proctable = OPAL_INT_TO_BOOL(value);
mca_base_param_reg_string_name("orte", "debugger_test_daemon",
"Name of the executable to be used to simulate a debugger colaunch (relative or absolute path)",
false, false, NULL, &orte_debugger_base.test_daemon);
mca_base_param_reg_int_name("orte",
"debugger_test_attach",
"Test debugger colaunch after debugger attachment",
false, false, 0, &value);
orte_debugger_base.test_attach = OPAL_INT_TO_BOOL(value);
/* Open up all available components */
if (ORTE_SUCCESS !=
mca_base_components_open("debugger", orte_debugger_base.output,
mca_debugger_base_static_components,
&orte_debugger_base_components_available,
true)) {
return ORTE_ERROR;
}
/* All done */
return ORTE_SUCCESS;
}
int mca_bml_r2_component_open(void)
{
int tmp;
mca_base_param_reg_int(&mca_bml_r2_component.bml_version,
"show_unreach_errors",
"Show error message when procs are unreachable",
false,
false,
1,
&tmp);
mca_bml_r2.show_unreach_errors = OPAL_INT_TO_BOOL(tmp);
return OMPI_SUCCESS;
}
int opal_cr_refresh_environ(int prev_pid) {
int val;
char *file_name = NULL;
struct stat file_status;
if( 0 >= prev_pid ) {
prev_pid = getpid();
}
/*
* Make sure the file exists. If it doesn't then this means 2 things:
* 1) We have already executed this function, and
* 2) The file has been deleted on the previous round.
*/
asprintf(&file_name, "%s/%s-%d", opal_tmp_directory(), OPAL_CR_BASE_ENV_NAME, prev_pid);
if(0 != stat(file_name, &file_status) ){
return OPAL_SUCCESS;
}
#if OPAL_ENABLE_CRDEBUG == 1
opal_unsetenv(mca_base_param_env_var("opal_cr_enable_crdebug"), &environ);
#endif
extract_env_vars(prev_pid, file_name);
#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] (refresh)\n",
(MPIR_debug_with_checkpoint ? "True": "False"));
#else
val = 0; /* Silence Compiler warning */
#endif
if( NULL != file_name ){
free(file_name);
file_name = NULL;
}
return OPAL_SUCCESS;
}
static int test_open(void)
{
int tmp;
mca_base_param_reg_int(&mca_paffinity_test_component.super.base_version, "bound",
"Whether or not to test as if externally bound (default=0: no)",
false, false, (int)false, &tmp);
mca_paffinity_test_component.bound = OPAL_INT_TO_BOOL(tmp);
mca_base_param_reg_int(&mca_paffinity_test_component.super.base_version, "num_sockets",
"Number of sockets on each node (default=4)",
false, false, 4, &mca_paffinity_test_component.num_sockets);
mca_base_param_reg_int(&mca_paffinity_test_component.super.base_version, "num_cores",
"Number of cores in each socket (default=4)",
false, false, 4, &mca_paffinity_test_component.num_cores);
return OPAL_SUCCESS;
}
int orte_proc_info(void)
{
int id, tmp;
/* all other params are set elsewhere */
id = mca_base_param_register_int("seed", NULL, NULL, NULL, orte_process_info.seed);
mca_base_param_lookup_int(id, &tmp);
orte_process_info.seed = OPAL_INT_TO_BOOL(tmp);
/* if we are a seed, then make sure the daemon flag is NOT set so that
* framework components are properly selected
*/
if (orte_process_info.seed) {
orte_process_info.daemon = false;
}
id = mca_base_param_register_int("orte", "app", "num", NULL, -1);
mca_base_param_lookup_int(id, &tmp);
orte_process_info.app_num = tmp;
id = mca_base_param_register_string("gpr", "replica", "uri", NULL, orte_process_info.gpr_replica_uri);
mca_base_param_lookup_string(id, &(orte_process_info.gpr_replica_uri));
mca_base_param_set_internal(id, true);
id = mca_base_param_register_string("ns", "replica", "uri", NULL, orte_process_info.ns_replica_uri);
mca_base_param_lookup_string(id, &(orte_process_info.ns_replica_uri));
mca_base_param_set_internal(id, true);
id = mca_base_param_register_string("tmpdir", "base", NULL, NULL, orte_process_info.tmpdir_base);
mca_base_param_lookup_string(id, &(orte_process_info.tmpdir_base));
/* get the process id */
orte_process_info.pid = getpid();
return ORTE_SUCCESS;
}
/*
* Start monitoring of local processes
*/
static void start(orte_jobid_t jobid)
{
mca_base_component_t *c = &mca_sensor_file_component.super.base_version;
opal_list_item_t *item;
orte_odls_job_t *jobdat;
orte_app_context_t *app, *aptr;
int rc, tmp;
char *filename;
file_tracker_t *ft;
/* cannot monitor my own job */
if (jobid == ORTE_PROC_MY_NAME->jobid && ORTE_JOBID_WILDCARD != jobid) {
return;
}
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s starting file monitoring for job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jobid)));
/* get the local jobdat for this job */
for (item = opal_list_get_first(&orte_local_jobdata);
item != opal_list_get_end(&orte_local_jobdata);
item = opal_list_get_end(&orte_local_jobdata)) {
jobdat = (orte_odls_job_t*)item;
if (jobid == jobdat->jobid || ORTE_JOBID_WILDCARD == jobid) {
/* must be at least one app_context, so use the first one found */
app = NULL;
for (tmp=0; tmp < jobdat->apps.size; tmp++) {
if (NULL != (aptr = (orte_app_context_t*)opal_pointer_array_get_item(&jobdat->apps, tmp))) {
app = aptr;
break;
}
}
if (NULL == app) {
/* got a problem */
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
continue;
}
/* search the environ to get the filename */
if (ORTE_SUCCESS != (rc = mca_base_param_find_string(c, "filename", app->env, &filename))) {
/* was a default file given */
if (NULL == mca_sensor_file_component.file) {
/* can't do anything without a file */
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s sensor:file no file for job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jobid)));
continue;
}
filename = mca_sensor_file_component.file;
}
/* create the tracking object */
ft = OBJ_NEW(file_tracker_t);
ft->jobid = jobid;
ft->file = strdup(filename);
/* search the environ to see what we are checking */
tmp = 0;
if (ORTE_SUCCESS != (rc = mca_base_param_find_int(c, "check_size", app->env, &tmp))) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_size) {
ft->check_size = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_size);
}
} else {
ft->check_size = OPAL_INT_TO_BOOL(tmp);
}
tmp = 0;
if (ORTE_SUCCESS != (rc = mca_base_param_find_int(c, "check_access", app->env, &tmp))) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_access) {
ft->check_access = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_access);
}
} else {
ft->check_access = OPAL_INT_TO_BOOL(tmp);
}
tmp = 0;
if (ORTE_SUCCESS != (rc = mca_base_param_find_int(c, "check_mod", app->env, &tmp))) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_mod) {
ft->check_mod = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_mod);
}
} else {
ft->check_mod = OPAL_INT_TO_BOOL(tmp);
}
tmp = 0;
if (ORTE_SUCCESS != (rc = mca_base_param_find_int(c, "limit", app->env, &tmp))) {
ft->limit = mca_sensor_file_component.limit;
} else {
ft->limit = tmp;
}
opal_list_append(&jobs, &ft->super);
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s file %s monitored for %s%s%s with limit %d",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ft->file, ft->check_size ? "SIZE:" : " ",
ft->check_access ? "ACCESS TIME:" : " ",
ft->check_mod ? "MOD TIME" : " ", ft->limit));
}
}
/* start sampling */
if (NULL == sample_ev && !opal_list_is_empty(&jobs)) {
/* startup a timer to wake us up periodically
* for a data sample
*/
sample_ev = (opal_event_t *) malloc(sizeof(opal_event_t));
opal_event_evtimer_set(opal_event_base, sample_ev, sample, sample_ev);
sample_time.tv_sec = mca_sensor_file_component.sample_rate;
sample_time.tv_usec = 0;
opal_event_evtimer_add(sample_ev, &sample_time);
}
return;
}
#include "opal/runtime/opal_params.h" #include "opal/dss/dss.h" #include "opal/util/show_help.h" #include "opal/util/timings.h" char *opal_signal_string = NULL; char *opal_net_private_ipv4 = NULL; char *opal_set_max_sys_limits = NULL; #if OPAL_ENABLE_TIMING char *opal_timing_sync_file = NULL; char *opal_timing_output = NULL; bool opal_timing_overhead = true; #endif bool opal_built_with_cuda_support = OPAL_INT_TO_BOOL(OPAL_CUDA_SUPPORT); bool opal_cuda_support = false; #if OPAL_ENABLE_FT_CR == 1 bool opal_base_distill_checkpoint_ready = false; #endif /** * Globals imported from the OMPI layer. */ int opal_leave_pinned = -1; bool opal_leave_pinned_pipeline = false; bool opal_abort_print_stack = false; int opal_abort_delay = 0; static bool opal_register_done = false;
* included in all MPI function imlementation files * * The values below are the default values. */ bool ompi_mpi_param_check = true; bool ompi_debug_show_handle_leaks = false; int ompi_debug_show_mpi_alloc_mem_leaks = 0; bool ompi_debug_no_free_handles = false; bool ompi_mpi_show_mca_params = false; char *ompi_mpi_show_mca_params_file = NULL; bool ompi_mpi_abort_print_stack = false; int ompi_mpi_abort_delay = 0; bool ompi_mpi_keep_fqdn_hostnames = false; int ompi_mpi_leave_pinned = -1; bool ompi_mpi_leave_pinned_pipeline = false; bool ompi_have_sparse_group_storage = OPAL_INT_TO_BOOL(OMPI_GROUP_SPARSE); bool ompi_use_sparse_group_storage = OPAL_INT_TO_BOOL(OMPI_GROUP_SPARSE); bool ompi_mpi_built_with_cuda_support = OPAL_INT_TO_BOOL(OMPI_CUDA_SUPPORT); bool ompi_mpi_cuda_support; uint32_t ompi_hostname_cutoff = UINT32_MAX; bool ompi_mpi_yield_when_idle = true; int ompi_mpi_event_tick_rate = -1; char *ompi_mpi_show_mca_params_string = NULL; bool ompi_mpi_have_sparse_group_storage = !!(OMPI_GROUP_SPARSE); bool ompi_mpi_preconnect_mpi = false; static bool show_default_mca_params = false; static bool show_file_mca_params = false; static bool show_enviro_mca_params = false; static bool show_override_mca_params = false;
/**
* Function for finding and opening either all MCA components, or the
* one that was specifically requested via a MCA parameter.
*/
int mca_base_components_open(const char *type_name, int output_id,
const mca_base_component_t **static_components,
opal_list_t *components_available,
bool open_dso_components)
{
int ret, param;
opal_list_item_t *item;
opal_list_t components_found;
char **requested_component_names;
int param_verbose = -1;
int param_type = -1;
int verbose_level;
char *str;
bool include_mode;
#if (OPAL_ENABLE_FT == 1) && (OPAL_ENABLE_FT_CR == 1)
opal_list_item_t *next;
uint32_t open_only_flags = MCA_BASE_METADATA_PARAM_NONE;
const mca_base_component_t *component;
#endif
/* Register MCA parameters */
/* Check to see if it exists first */
if( 0 > (param_type = mca_base_param_find(type_name, NULL, NULL) ) ) {
asprintf(&str, "Default selection set of components for the %s framework (<none>"
" means use all components that can be found)", type_name);
param_type =
mca_base_param_reg_string_name(type_name, NULL, str,
false, false, NULL, NULL);
free(str);
}
param = mca_base_param_find("mca", NULL, "component_show_load_errors");
mca_base_param_lookup_int(param, &ret);
show_errors = OPAL_INT_TO_BOOL(ret);
/* Setup verbosity for this MCA type */
asprintf(&str, "Verbosity level for the %s framework (0 = no verbosity)", type_name);
param_verbose =
mca_base_param_reg_int_name(type_name, "base_verbose",
str, false, false, 0, NULL);
free(str);
mca_base_param_lookup_int(param_verbose, &verbose_level);
if (output_id != 0) {
opal_output_set_verbosity(output_id, verbose_level);
}
opal_output_verbose(10, output_id,
"mca: base: components_open: Looking for %s components",
type_name);
ret = parse_requested(param_type, &include_mode, &requested_component_names);
if( OPAL_SUCCESS != ret ) {
return ret;
}
/* Find and load requested components */
if (OPAL_SUCCESS != (ret =
mca_base_component_find(NULL, type_name, static_components,
requested_component_names, include_mode,
&components_found, open_dso_components)) ) {
return ret;
}
#if (OPAL_ENABLE_FT == 1) && (OPAL_ENABLE_FT_CR == 1)
{
int param_id = -1;
int param_val = 0;
/*
* Extract supported mca parameters for selection contraints
* Supported Options:
* - mca_base_component_distill_checkpoint_ready = Checkpoint Ready
*/
param_id = mca_base_param_reg_int_name("mca", "base_component_distill_checkpoint_ready",
"Distill only those components that are Checkpoint Ready",
false, false,
0, ¶m_val);
if( 0 != param_val ) { /* Select Checkpoint Ready */
open_only_flags |= MCA_BASE_METADATA_PARAM_CHECKPOINT;
}
}
#endif /* (OPAL_ENABLE_FT == 1) && (OPAL_ENABLE_FT_CR == 1) */
/*
* Pre-process the list with parameter constraints
* e.g., If requested to select only CR enabled components
* then only make available those components.
*
* JJH Note: Currently checkpoint/restart is the only user of this
* functionality. If other component constraint options are
* added, then this logic can be used for all contraint
* options.
*/
#if (OPAL_ENABLE_FT == 1) && (OPAL_ENABLE_FT_CR == 1)
if( !(MCA_BASE_METADATA_PARAM_NONE & open_only_flags) ) {
if( MCA_BASE_METADATA_PARAM_CHECKPOINT & open_only_flags) {
opal_output_verbose(10, output_id,
"mca: base: components_open: "
"including only %s components that are checkpoint enabled", type_name);
}
/*
* Check all the components to make sure they adhere to the user
* expressed requirements.
*/
for(item = opal_list_get_first(&components_found);
item != opal_list_get_end(&components_found);
item = next ) {
mca_base_open_only_dummy_component_t *dummy;
mca_base_component_list_item_t *cli = (mca_base_component_list_item_t *) item;
dummy = (mca_base_open_only_dummy_component_t*) cli->cli_component;
component = cli->cli_component;
next = opal_list_get_next(item);
/*
* If the user asked for a checkpoint enabled run
* then only load checkpoint enabled components.
*/
if( MCA_BASE_METADATA_PARAM_CHECKPOINT & open_only_flags) {
if( MCA_BASE_METADATA_PARAM_CHECKPOINT & dummy->data.param_field) {
opal_output_verbose(10, output_id,
"mca: base: components_open: "
"(%s) Component %s is Checkpointable",
type_name,
dummy->version.mca_component_name);
}
else {
opal_output_verbose(10, output_id,
"mca: base: components_open: "
"(%s) Component %s is *NOT* Checkpointable - Disabled",
type_name,
dummy->version.mca_component_name);
opal_list_remove_item(&components_found, item);
/* Make sure to release the component since we are not
* opening it */
mca_base_component_repository_release(component);
}
}
}
}
#endif /* (OPAL_ENABLE_FT == 1) && (OPAL_ENABLE_FT_CR == 1) */
/* Open all remaining components */
ret = open_components(type_name, output_id,
&components_found, components_available);
/* Free resources */
for (item = opal_list_remove_first(&components_found); NULL != item;
item = opal_list_remove_first(&components_found)) {
OBJ_RELEASE(item);
}
OBJ_DESTRUCT(&components_found);
if (NULL != requested_component_names) {
opal_argv_free(requested_component_names);
}
/* All done */
return ret;
}
int ompi_mpi_register_params(void)
{
int value;
char *param;
/* Whether we want MPI API function parameter checking or not */
mca_base_param_reg_int_name("mpi", "param_check",
"Whether you want MPI API parameters checked at run-time or not. Possible values are 0 (no checking) and 1 (perform checking at run-time)",
false, false, MPI_PARAM_CHECK, &value);
ompi_mpi_param_check = OPAL_INT_TO_BOOL(value);
if (ompi_mpi_param_check) {
value = 0;
if (MPI_PARAM_CHECK) {
value = 1;
}
if (0 == value) {
orte_show_help("help-mpi-runtime.txt",
"mpi-param-check-enabled-but-compiled-out",
true);
ompi_mpi_param_check = false;
}
}
/*
* opal_progress: decide whether to yield and the event library
* tick rate
*/
/* JMS: Need ORTE data here -- set this to 0 when
exactly/under-subscribed, or 1 when oversubscribed */
mca_base_param_reg_int_name("mpi", "yield_when_idle",
"Yield the processor when waiting for MPI communication (for MPI processes, will default to 1 when oversubscribing nodes)",
false, false, -1, NULL);
mca_base_param_reg_int_name("mpi", "event_tick_rate",
"How often to progress TCP communications (0 = never, otherwise specified in microseconds)",
false, false, -1, NULL);
/* Whether or not to show MPI handle leaks */
mca_base_param_reg_int_name("mpi", "show_handle_leaks",
"Whether MPI_FINALIZE shows all MPI handles that were not freed or not",
false, false,
(int) ompi_debug_show_handle_leaks, &value);
ompi_debug_show_handle_leaks = OPAL_INT_TO_BOOL(value);
/* Whether or not to free MPI handles. Useless without run-time
param checking, so implicitly set that to true if we don't want
to free the handles. */
mca_base_param_reg_int_name("mpi", "no_free_handles",
"Whether to actually free MPI objects when their handles are freed",
false, false,
(int) ompi_debug_no_free_handles, &value);
ompi_debug_no_free_handles = OPAL_INT_TO_BOOL(value);
if (ompi_debug_no_free_handles) {
ompi_mpi_param_check = true;
value = 0;
if (MPI_PARAM_CHECK) {
value = 1;
}
if (0 == value) {
opal_output(0, "WARNING: MCA parameter mpi_no_free_handles set to true, but MPI");
opal_output(0, "WARNING: parameter checking has been compiled out of Open MPI.");
opal_output(0, "WARNING: mpi_no_free_handles is therefore only partially effective!");
}
}
/* Whether or not to show MPI_ALLOC_MEM leaks */
mca_base_param_reg_int_name("mpi", "show_mpi_alloc_mem_leaks",
"If >0, MPI_FINALIZE will show up to this many instances of memory allocated by MPI_ALLOC_MEM that was not freed by MPI_FREE_MEM",
false, false,
ompi_debug_show_mpi_alloc_mem_leaks,
&ompi_debug_show_mpi_alloc_mem_leaks);
/* Whether or not to print all MCA parameters in MPI_INIT */
mca_base_param_reg_string_name("mpi", "show_mca_params",
"Whether to show all MCA parameter values during MPI_INIT or not (good for reproducability of MPI jobs "
"for debug purposes). Accepted values are all, default, file, api, and enviro - or a comma "
"delimited combination of them",
false, false, NULL, ¶m);
if (NULL != param) {
char **args;
int i;
ompi_mpi_show_mca_params = true;
args = opal_argv_split(param, ',');
if (NULL == args) {
opal_output(0, "WARNING: could not parse mpi_show_mca_params request - defaulting to show \"all\"");
show_default_mca_params = true;
show_file_mca_params = true;
show_enviro_mca_params = true;
show_override_mca_params = true;
} else {
for (i=0; NULL != args[i]; i++) {
if (0 == strcasecmp(args[i], "all") || 0 == strcmp(args[i], "1")) {
show_default_mca_params = true;
show_file_mca_params = true;
show_enviro_mca_params = true;
show_override_mca_params = true;
} else if (0 == strcasecmp(args[i], "default")) {
show_default_mca_params = true;
} else if (0 == strcasecmp(args[i], "file")) {
show_file_mca_params = true;
} else if (0 == strcasecmp(args[i], "enviro") ||
0 == strcasecmp(args[i], "env")) {
show_enviro_mca_params = true;
} else if (0 == strcasecmp(args[i], "api")) {
show_override_mca_params = true;
}
}
opal_argv_free(args);
}
}
/* File to use when dumping the parameters */
mca_base_param_reg_string_name("mpi", "show_mca_params_file",
"If mpi_show_mca_params is true, setting this string to a valid filename tells Open MPI to dump all the MCA parameter values into a file suitable for reading via the mca_param_files parameter (good for reproducability of MPI jobs)",
false, false,
"", &ompi_mpi_show_mca_params_file);
/* User-level process pinning controls */
/* Do we want to save hostnames for debugging messages? This can
eat quite a bit of memory... */
mca_base_param_reg_int_name("mpi", "keep_peer_hostnames",
"If nonzero, save the string hostnames of all MPI peer processes (mostly for error / debugging output messages). This can add quite a bit of memory usage to each MPI process.",
false, false, 1, &value);
ompi_mpi_keep_peer_hostnames = OPAL_INT_TO_BOOL(value);
/* MPI_ABORT controls */
mca_base_param_reg_int_name("mpi", "abort_delay",
"If nonzero, print out an identifying message when MPI_ABORT is invoked (hostname, PID of the process that called MPI_ABORT) and delay for that many seconds before exiting (a negative delay value means to never abort). This allows attaching of a debugger before quitting the job.",
false, false,
ompi_mpi_abort_delay,
&ompi_mpi_abort_delay);
mca_base_param_reg_int_name("mpi", "abort_print_stack",
"If nonzero, print out a stack trace when MPI_ABORT is invoked",
false,
/* If we do not have stack trace
capability, make this a read-only
MCA param */
#if OMPI_WANT_PRETTY_PRINT_STACKTRACE && ! defined(__WINDOWS__) && defined(HAVE_BACKTRACE)
false,
#else
true,
#endif
(int) ompi_mpi_abort_print_stack,
&value);
#if OMPI_WANT_PRETTY_PRINT_STACKTRACE && ! defined(__WINDOWS__) && defined(HAVE_BACKTRACE)
/* Only take the value if we have stack trace capability */
ompi_mpi_abort_print_stack = OPAL_INT_TO_BOOL(value);
#else
/* If we do not have stack trace capability, ensure that this is
hard-coded to false */
ompi_mpi_abort_print_stack = false;
#endif
value = mca_base_param_reg_int_name("mpi", "preconnect_mpi",
"Whether to force MPI processes to fully "
"wire-up the MPI connections between MPI "
"processes during "
"MPI_INIT (vs. making connections lazily -- "
"upon the first MPI traffic between each "
"process peer pair)",
false, false, 0, NULL);
mca_base_param_reg_syn_name(value, "mpi", "preconnect_all", true);
/* Leave pinned parameter */
mca_base_param_reg_int_name("mpi", "leave_pinned",
"Whether to use the \"leave pinned\" protocol or not. Enabling this setting can help bandwidth performance when repeatedly sending and receiving large messages with the same buffers over RDMA-based networks (0 = do not use \"leave pinned\" protocol, 1 = use \"leave pinned\" protocol, -1 = allow network to choose at runtime).",
false, false,
ompi_mpi_leave_pinned, &value);
ompi_mpi_leave_pinned = (value >= 1) ? true: false;
mca_base_param_reg_int_name("mpi", "leave_pinned_pipeline",
"Whether to use the \"leave pinned pipeline\" protocol or not.",
false, false,
(int) ompi_mpi_leave_pinned_pipeline, &value);
ompi_mpi_leave_pinned_pipeline = OPAL_INT_TO_BOOL(value);
if (ompi_mpi_leave_pinned && ompi_mpi_leave_pinned_pipeline) {
ompi_mpi_leave_pinned_pipeline = 0;
orte_show_help("help-mpi-runtime.txt",
"mpi-params:leave-pinned-and-pipeline-selected",
true);
}
mca_base_param_reg_int_name("mpi", "warn_on_fork",
"If nonzero, issue a warning if program forks under conditions that could cause system errors",
false, false,
(int) true, &value);
ompi_warn_on_fork = OPAL_INT_TO_BOOL(value);
/* Sparse group storage support */
mca_base_param_reg_int_name("mpi", "have_sparse_group_storage",
"Whether this Open MPI installation supports storing of data in MPI groups in \"sparse\" formats (good for extremely large process count MPI jobs that create many communicators/groups)",
false, true, (int) OMPI_GROUP_SPARSE, NULL);
mca_base_param_reg_int_name("mpi", "use_sparse_group_storage",
"Whether to use \"sparse\" storage formats for MPI groups (only relevant if mpi_have_sparse_group_storage is 1)",
false, false, OMPI_GROUP_SPARSE, &value);
ompi_use_sparse_group_storage = OPAL_INT_TO_BOOL(value);
if (ompi_use_sparse_group_storage) {
value = 0;
if (OMPI_GROUP_SPARSE) {
value = 1;
}
if (0 == value) {
orte_show_help("help-mpi-runtime.txt",
"sparse groups enabled but compiled out",
true);
ompi_use_sparse_group_storage = false;
}
}
/* The ddt engine has a few parameters */
return ompi_ddt_register_params();
}
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_register_params(void)
{
int ret;
/*
* This string is going to be used in opal/util/stacktrace.c
*/
{
char *string = NULL;
int j;
int signals[] = {
#ifdef SIGABRT
SIGABRT,
#endif
#ifdef SIGBUS
SIGBUS,
#endif
#ifdef SIGFPE
SIGFPE,
#endif
#ifdef SIGSEGV
SIGSEGV,
#endif
-1
};
for (j = 0 ; signals[j] != -1 ; ++j) {
if (j == 0) {
asprintf(&string, "%d", signals[j]);
} else {
char *tmp;
asprintf(&tmp, "%s,%d", string, signals[j]);
free(string);
string = tmp;
}
}
mca_base_param_reg_string_name("opal", "signal",
"Comma-delimited list of integer signal numbers to Open MPI to attempt to intercept. Upon receipt of the intercepted signal, Open MPI will display a stack trace and abort. Open MPI will *not* replace signals if handlers are already installed by the time MPI_INIT is invoked. Optionally append \":complain\" to any signal number in the comma-delimited list to make Open MPI complain if it detects another signal handler (and therefore does not insert its own).",
false, false, string, NULL);
free(string);
}
{
int j;
mca_base_param_reg_int_name("opal", "profile",
"Set to non-zero to profile component selections",
false, false, (int)false, &j);
opal_profile = OPAL_INT_TO_BOOL(j);
mca_base_param_reg_string_name("opal", "profile_file",
"Name of the file containing the cluster configuration information",
false, false, NULL, &opal_profile_file);
}
#if OPAL_ENABLE_DEBUG
mca_base_param_reg_int_name("opal", "progress_debug",
"Set to non-zero to debug progress engine features",
false, false, 0, NULL);
{
int value;
mca_base_param_reg_int_name("opal", "debug_locks",
"Debug mutex usage within Open MPI. On a "
"non-threaded build, this enables integer counters and "
"warning messages when double-locks are detected.",
false, false, 0, &value);
if (value) opal_mutex_check_locks = true;
}
#endif
/* The ddt engine has a few parameters */
ret = opal_datatype_register_params();
if (OPAL_SUCCESS != ret) {
return ret;
}
/* Paffinity base also has some parameters */
return opal_paffinity_base_register_params();
}
int opal_register_params(void)
{
int value;
/*
* This string is going to be used in opal/util/stacktrace.c
*/
{
char *string = NULL;
int j;
int signals[] = {
#ifdef SIGABRT
SIGABRT,
#endif
#ifdef SIGBUS
SIGBUS,
#endif
#ifdef SIGFPE
SIGFPE,
#endif
#ifdef SIGSEGV
SIGSEGV,
#endif
-1
};
for (j = 0 ; signals[j] != -1 ; ++j) {
if (j == 0) {
asprintf(&string, "%d", signals[j]);
} else {
char *tmp;
asprintf(&tmp, "%s,%d", string, signals[j]);
free(string);
string = tmp;
}
}
mca_base_param_reg_string_name("opal", "signal",
"Comma-delimited list of integer signal numbers to Open MPI to attempt to intercept. Upon receipt of the intercepted signal, Open MPI will display a stack trace and abort. Open MPI will *not* replace signals if handlers are already installed by the time MPI_INIT is invoked. Optionally append \":complain\" to any signal number in the comma-delimited list to make Open MPI complain if it detects another signal handler (and therefore does not insert its own).",
false, false, string, NULL);
free(string);
}
#if OMPI_ENABLE_DEBUG
mca_base_param_reg_int_name("opal", "progress_debug",
"Set to non-zero to debug progress engine features",
false, false, 0, NULL);
{
mca_base_param_reg_int_name("opal", "debug_locks",
"Debug mutex usage within Open MPI. On a "
"non-threaded build, this enables integer counters and "
"warning messages when double-locks are detected.",
false, false, 0, &value);
if (value) opal_mutex_check_locks = true;
}
#endif
/*
* Do we want the "warning: your mmap file is on NFS!" message? Per a
* thread on the OMPI devel list
* (http://www.open-mpi.org/community/lists/devel/2011/12/10054.php),
* on some systems, it doesn't seem to matter. But per older threads,
* it definitely does matter on some systems. Perhaps newer kernels
* are smarter about this kind of stuff...? Regardless, we should
* provide the ability to turn off this message for systems where the
* effect doesn't matter.
*
* v1.4.x-specific note: the MCA param name is "shmem_mmap_...",
* where "shmem" is not a framework that exists in the v1.4
* series. This parameter was added right before 1.4.5, and at a
* similar time as 1.5.5 (where the "shmem" framework *does*
* exist). The idea was to have a consistent MCA param name
* starting with v1.4.5. Hence, we put a slightly non-sensiscal
* name here in v1.4.x so that we'd have a correct/good name
* moving forward.
*/
mca_base_param_reg_int_name("shmem",
"mmap_enable_nfs_warning",
"Enable the warning emitted when Open MPI detects that its shared memory backing file is located on a network filesystem (1 = enabled, 0 = disabled).",
false, false,
(int)true, &value);
opal_mmap_on_nfs_warning = OPAL_INT_TO_BOOL(value);
/* Paffinity base also has some parameters */
return opal_paffinity_base_register_params();
}
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;
}
int orcm_init(orcm_proc_type_t flags)
{
int ret;
char *error;
int i, spin;
if (NULL != getenv("ORCM_MCA_spin")) {
spin = 1;
/* spin until a debugger can attach */
while (0 != spin) {
ret = 0;
while (ret < 10000) {
ret++;
};
}
}
if (!orcm_util_initialized) {
orcm_init_util();
}
/* set the default leader policy */
orcm_default_leader_policy.jobid = ORTE_JOBID_WILDCARD;
orcm_default_leader_policy.vpid = ORTE_VPID_WILDCARD;
/* get the number of max msgs */
mca_base_param_reg_int_name("orcm", "max_buffered_msgs",
"Number of recvd messages to hold in storage from each source",
false, false, ORCM_MAX_MSG_RING_SIZE, &orcm_max_msg_ring_size);
/* independent mode or not */
mca_base_param_reg_int_name("orcm", "sched_kill_dvm",
"Whether or not scheduler kills associated daemons upon termination (default: no)",
false, false, (int)false, &ret);
orcm_sched_kill_dvm = OPAL_INT_TO_BOOL(ret);
/* setup the globals that require initialization */
orcm_triplets = OBJ_NEW(orcm_triplets_array_t);
#ifdef HAVE_QSYSTEM_H
#ifdef Q_SYSTEM_INTFCS_TO_PROBE_FOR_IP_ADDRESS
{
char *eth_ifs[] = Q_SYSTEM_INTFCS_TO_PROBE_FOR_IP_ADDRESS;
char **adds=NULL, *ifs, *envar;
int i, num_ifs;
num_ifs = sizeof(eth_ifs) / sizeof(eth_ifs[0]);
for (i=0; i < num_ifs; i++) {
opal_argv_append_nosize(&adds, eth_ifs[i]);
}
ifs = opal_argv_join(adds, ',');
opal_argv_free(adds);
/* push it into the environ so that the rmcast framework can get it */
asprintf(&envar, "OMPI_MCA_rmcast_base_if_include=%s", ifs);
putenv(envar);
/* cannot release envar as the environ doesn't keep its own copy */
free(ifs);
}
#endif
#endif
/* initialize us */
if (ORTE_SUCCESS != (ret = orte_init(NULL, NULL, flags))) {
error = "orte_init";
goto error;
}
if (!ORCM_PROC_IS_TOOL) {
opal_set_using_threads(true);
}
if (!ORCM_PROC_IS_APP) {
trap_signals();
}
orcm_initialized = true;
return ORCM_SUCCESS;
error:
if (ORCM_ERR_SILENT != ret) {
orte_show_help("help-openrcm-runtime.txt",
"orcm_init:startup:internal-failure",
true, error, ORTE_ERROR_NAME(ret), ret);
}
return ret;
}
/*
* Start monitoring of local processes
*/
static void start(orte_jobid_t jobid)
{
orte_job_t *jobdat;
orte_app_context_t *app, *aptr;
int i;
char *filename;
file_tracker_t *ft;
char *ptr;
/* cannot monitor my own job */
if (jobid == ORTE_PROC_MY_NAME->jobid && ORTE_JOBID_WILDCARD != jobid) {
return;
}
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s starting file monitoring for job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jobid)));
/* get the local jobdat for this job */
if (NULL == (jobdat = orte_get_job_data_object(jobid))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
return;
}
/* must be at least one app_context, so use the first one found */
app = NULL;
for (i=0; i < jobdat->apps->size; i++) {
if (NULL != (aptr = (orte_app_context_t*)opal_pointer_array_get_item(jobdat->apps, i))) {
app = aptr;
break;
}
}
if (NULL == app) {
/* got a problem */
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
return;
}
/* search the environ to get the filename */
if (!find_value(app, "OMPI_MCA_sensor_file_filename", &filename)) {
/* was a default file given */
if (NULL == mca_sensor_file_component.file) {
/* can't do anything without a file */
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s sensor:file no file for job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jobid)));
return;
}
filename = mca_sensor_file_component.file;
}
/* create the tracking object */
ft = OBJ_NEW(file_tracker_t);
ft->jobid = jobid;
ft->file = strdup(filename);
/* search the environ to see what we are checking */
if (!find_value(app, "OMPI_MCA_sensor_file_check_size", &ptr)) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_size) {
ft->check_size = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_size);
}
} else {
ft->check_size = OPAL_INT_TO_BOOL(strtol(ptr, NULL, 10));
free(ptr);
}
if (!find_value(app, "OMPI_MCA_sensor_file_check_access", &ptr)) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_access) {
ft->check_access = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_access);
}
} else {
ft->check_access = OPAL_INT_TO_BOOL(strtol(ptr, NULL, 10));
free(ptr);
}
if (!find_value(app, "OMPI_MCA_sensor_file_check_mod", &ptr)) {
/* was a default value given */
if (0 < mca_sensor_file_component.check_mod) {
ft->check_mod = OPAL_INT_TO_BOOL(mca_sensor_file_component.check_mod);
}
} else {
ft->check_mod = OPAL_INT_TO_BOOL(strtol(ptr, NULL, 10));
free(ptr);
}
if (!find_value(app, "OMPI_MCA_sensor_file_limit", &ptr)) {
ft->limit = mca_sensor_file_component.limit;
} else {
ft->limit = strtol(ptr, NULL, 10);
free(ptr);
}
opal_list_append(&jobs, &ft->super);
OPAL_OUTPUT_VERBOSE((1, orte_sensor_base.output,
"%s file %s monitored for %s%s%s with limit %d",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ft->file, ft->check_size ? "SIZE:" : " ",
ft->check_access ? "ACCESS TIME:" : " ",
ft->check_mod ? "MOD TIME" : " ", ft->limit));
return;
}
void mca_common_cuda_init(void)
{
int id, value;
CUresult res;
CUcontext cuContext;
if (initialized) {
return;
}
/* Set different levels of verbosity in the cuda related code. */
id = mca_base_param_reg_int_name("mpi", "common_cuda_verbose",
"Set level of common cuda verbosity",
false, false, 0, &mca_common_cuda_verbose);
mca_common_cuda_output = opal_output_open(NULL);
opal_output_set_verbosity(mca_common_cuda_output, mca_common_cuda_verbose);
/* Control whether system buffers get CUDA pinned or not. Allows for
* performance analysis. */
id = mca_base_param_reg_int_name("mpi", "common_cuda_register_memory",
"Whether to cuMemHostRegister preallocated BTL buffers",
false, false,
(int) mca_common_cuda_register_memory, &value);
mca_common_cuda_register_memory = OPAL_INT_TO_BOOL(value);
/* Control whether we see warnings when CUDA memory registration fails. This is
* useful when CUDA support is configured in, but we are running a regular MPI
* application without CUDA. */
id = mca_base_param_reg_int_name("mpi", "common_cuda_warning",
"Whether to print warnings when CUDA registration fails",
false, false,
(int) mca_common_cuda_warning, &value);
mca_common_cuda_warning = OPAL_INT_TO_BOOL(value);
/* Check to see if this process is running in a CUDA context. If
* so, all is good. If not, then disable registration of memory. */
res = cuCtxGetCurrent(&cuContext);
if (CUDA_SUCCESS != res) {
if (mca_common_cuda_warning) {
/* Check for the not initialized error since we can make suggestions to
* user for this error. */
if (CUDA_ERROR_NOT_INITIALIZED == res) {
orte_show_help("help-mpi-common-cuda.txt", "cuCtxGetCurrent failed not initialized",
true);
} else {
orte_show_help("help-mpi-common-cuda.txt", "cuCtxGetCurrent failed",
true, res);
}
}
mca_common_cuda_enabled = false;
mca_common_cuda_register_memory = false;
} else if ((CUDA_SUCCESS == res) && (NULL == cuContext)) {
if (mca_common_cuda_warning) {
orte_show_help("help-mpi-common-cuda.txt", "cuCtxGetCurrent returned NULL",
true);
}
mca_common_cuda_enabled = false;
mca_common_cuda_register_memory = false;
} else {
/* All is good. mca_common_cuda_register_memory will retain its original
* value. Normally, that is 1, but the user can override it to disable
* registration of the internal buffers. */
mca_common_cuda_enabled = true;
opal_output_verbose(20, mca_common_cuda_output,
"CUDA: cuCtxGetCurrent succeeded");
}
opal_output_verbose(30, mca_common_cuda_output,
"CUDA: initialized");
initialized = true;
}
int orte_register_params(void)
{
int id;
opal_output_stream_t lds;
/* only go thru this once - mpirun calls it twice, which causes
* any error messages to show up twice
*/
if (passed_thru) {
return ORTE_SUCCESS;
}
passed_thru = true;
/* get a clean output channel too - need to do this here because
* we use it below, and orterun and some other tools call this
* function prior to calling orte_init
*/
OBJ_CONSTRUCT(&lds, opal_output_stream_t);
lds.lds_want_stdout = true;
orte_clean_output = opal_output_open(&lds);
OBJ_DESTRUCT(&lds);
orte_help_want_aggregate = true;
(void) mca_base_var_register ("orte", "orte", "base", "help_aggregate",
"If orte_base_help_aggregate is true, duplicate help messages will be aggregated rather than displayed individually. This can be helpful for parallel jobs that experience multiple identical failures; rather than print out the same help/failure message N times, display it once with a count of how many processes sent the same message.",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, MCA_BASE_VAR_FLAG_SETTABLE,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_help_want_aggregate);
/* LOOK FOR A TMP DIRECTORY BASE */
/* Several options are provided to cover a range of possibilities:
*
* (a) all processes need to use a specified location as the base
* for tmp directories
* (b) daemons on remote nodes need to use a specified location, but
* one different from that used by mpirun
* (c) mpirun needs to use a specified location, but one different
* from that used on remote nodes
*/
orte_tmpdir_base = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "tmpdir_base",
"Base of the session directory tree to be used by all processes",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_tmpdir_base);
orte_local_tmpdir_base = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "local_tmpdir_base",
"Base of the session directory tree to be used by orterun/mpirun",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_local_tmpdir_base);
orte_remote_tmpdir_base = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "remote_tmpdir_base",
"Base of the session directory tree on remote nodes, if required to be different from head node",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_remote_tmpdir_base);
/* if a global tmpdir was specified, then we do not allow specification
* of the local or remote values to avoid confusion
*/
if (NULL != orte_tmpdir_base &&
(NULL != orte_local_tmpdir_base || NULL != orte_remote_tmpdir_base)) {
opal_output(orte_clean_output,
"------------------------------------------------------------------\n"
"The MCA param orte_tmpdir_base was specified, which sets the base\n"
"of the temporary directory tree for all procs. However, values for\n"
"the local and/or remote tmpdir base were also given. This can lead\n"
"to confusion and is therefore not allowed. Please specify either a\n"
"global tmpdir base OR a local/remote tmpdir base value\n"
"------------------------------------------------------------------");
exit(1);
}
if (NULL != orte_tmpdir_base) {
if (NULL != orte_process_info.tmpdir_base) {
free(orte_process_info.tmpdir_base);
}
orte_process_info.tmpdir_base = strdup (orte_tmpdir_base);
} else if (ORTE_PROC_IS_HNP && NULL != orte_local_tmpdir_base) {
/* orterun will pickup the value for its own use */
if (NULL != orte_process_info.tmpdir_base) {
free(orte_process_info.tmpdir_base);
}
orte_process_info.tmpdir_base = strdup (orte_local_tmpdir_base);
} else if (ORTE_PROC_IS_DAEMON && NULL != orte_remote_tmpdir_base) {
/* orterun will pickup the value and forward it along, but must not
* use it in its own work. So only a daemon needs to get it, and the
* daemon will pass it down to its application procs. Note that orterun
* will pass -its- value to any procs local to it
*/
if (NULL != orte_process_info.tmpdir_base) {
free(orte_process_info.tmpdir_base);
}
orte_process_info.tmpdir_base = strdup (orte_remote_tmpdir_base);
}
orte_top_session_dir = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "top_session_dir",
"Top of the session directory tree for applications",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_top_session_dir);
if (NULL != orte_top_session_dir) {
if (NULL != orte_process_info.top_session_dir) {
free(orte_process_info.top_session_dir);
}
orte_process_info.top_session_dir = strdup(orte_top_session_dir);
}
orte_jobfam_session_dir = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "jobfam_session_dir",
"The jobfamily session directory for applications",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL_EQ,
&orte_jobfam_session_dir);
if (NULL != orte_jobfam_session_dir) {
if (NULL != orte_process_info.jobfam_session_dir) {
free(orte_process_info.jobfam_session_dir);
}
orte_process_info.jobfam_session_dir = strdup(orte_jobfam_session_dir);
}
orte_prohibited_session_dirs = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "no_session_dirs",
"Prohibited locations for session directories (multiple locations separated by ',', default=NULL)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_prohibited_session_dirs);
orte_create_session_dirs = true;
(void) mca_base_var_register ("orte", "orte", NULL, "create_session_dirs",
"Create session directories",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_create_session_dirs);
orte_execute_quiet = false;
(void) mca_base_var_register ("orte", "orte", NULL, "execute_quiet",
"Do not output error and help messages",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_execute_quiet);
orte_report_silent_errors = false;
(void) mca_base_var_register ("orte", "orte", NULL, "report_silent_errors",
"Report all errors, including silent ones",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_report_silent_errors);
orte_debug_flag = false;
(void) mca_base_var_register ("orte", "orte", NULL, "debug",
"Top-level ORTE debug switch (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_debug_flag);
orte_debug_verbosity = -1;
(void) mca_base_var_register ("orte", "orte", NULL, "debug_verbose",
"Verbosity level for ORTE debug messages (default: 1)",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_debug_verbosity);
orte_debug_daemons_file_flag = false;
(void) mca_base_var_register ("orte", "orte", NULL, "debug_daemons_file",
"Whether want stdout/stderr of daemons to go to a file or not",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_debug_daemons_file_flag);
/* If --debug-daemons-file was specified, that also implies
--debug-daemons */
if (orte_debug_daemons_file_flag) {
orte_debug_daemons_flag = true;
/* value can't change */
(void) mca_base_var_register ("orte", "orte", NULL, "debug_daemons",
"Whether to debug the ORTE daemons or not",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_CONSTANT,
&orte_debug_daemons_flag);
} else {
orte_debug_daemons_flag = false;
(void) mca_base_var_register ("orte", "orte", NULL, "debug_daemons",
"Whether to debug the ORTE daemons or not",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_debug_daemons_flag);
}
orte_progress_thread_debug_level = -1;
(void) mca_base_var_register ("orte", "orte", NULL, "progress_thread_debug",
"Debug level for ORTE progress threads",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_progress_thread_debug_level);
if (0 <= orte_progress_thread_debug_level) {
orte_progress_thread_debug = opal_output_open(NULL);
opal_output_set_verbosity(orte_progress_thread_debug,
orte_progress_thread_debug_level);
}
/* do we want session output left open? */
orte_leave_session_attached = false;
(void) mca_base_var_register ("orte", "orte", NULL, "leave_session_attached",
"Whether applications and/or daemons should leave their sessions "
"attached so that any output can be received - this allows X forwarding "
"without all the attendant debugging output",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_leave_session_attached);
/* if any debug level is set, ensure we output debug level dumps */
if (orte_debug_flag || orte_debug_daemons_flag || orte_leave_session_attached) {
orte_devel_level_output = true;
}
/* See comment in orte/tools/orterun/orterun.c about this MCA
param (this param is internal) */
orte_in_parallel_debugger = false;
(void) mca_base_var_register ("orte", "orte", NULL, "in_parallel_debugger",
"Whether the application is being debugged "
"in a parallel debugger (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, MCA_BASE_VAR_FLAG_INTERNAL,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_in_parallel_debugger);
orte_debugger_dump_proctable = false;
(void) mca_base_var_register ("orte", "orte", NULL, "output_debugger_proctable",
"Whether or not to output the debugger proctable after launch (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_ALL,
&orte_debugger_dump_proctable);
orte_debugger_test_daemon = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "debugger_test_daemon",
"Name of the executable to be used to simulate a debugger colaunch (relative or absolute path)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_debugger_test_daemon);
orte_debugger_test_attach = false;
(void) mca_base_var_register ("orte", "orte", NULL, "debugger_test_attach",
"Test debugger colaunch after debugger attachment",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_debugger_test_daemon);
orte_debugger_check_rate = 0;
(void) mca_base_var_register ("orte", "orte", NULL, "debugger_check_rate",
"Set rate (in secs) for auto-detect of debugger attachment (0 => do not check)",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_debugger_check_rate);
orte_do_not_launch = false;
(void) mca_base_var_register ("orte", "orte", NULL, "do_not_launch",
"Perform all necessary operations to prepare to launch the application, but do not actually launch it",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_do_not_launch);
orted_spin_flag = false;
(void) mca_base_var_register ("orte", "orte", NULL, "daemon_spin",
"Have any orteds spin until we can connect a debugger to them",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orted_spin_flag);
orted_debug_failure = ORTE_VPID_INVALID;
(void) mca_base_var_register ("orte", "orte", NULL, "daemon_fail",
"Have the specified orted fail after init for debugging purposes",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orted_debug_failure);
orted_debug_failure_delay = 0;
(void) mca_base_var_register ("orte", "orte", NULL, "daemon_fail_delay",
"Have the specified orted fail after specified number of seconds (default: 0 => no delay)",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orted_debug_failure_delay);
orte_startup_timeout = 0;
(void) mca_base_var_register ("orte", "orte", NULL, "startup_timeout",
"Seconds to wait for startup or job launch before declaring failed_to_start (default: 0 => do not check)",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_startup_timeout);
/* User-level debugger info string */
orte_base_user_debugger = "totalview @mpirun@ -a @mpirun_args@ : ddt -n @np@ -start @executable@ @executable_argv@ @single_app@ : fxp @mpirun@ -a @mpirun_args@";
(void) mca_base_var_register ("orte", "orte", NULL, "base_user_debugger",
"Sequence of user-level debuggers to search for in orterun",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_base_user_debugger);
#if 0
mca_base_param_reg_int_name("orte", "abort_timeout",
"Max time to wait [in secs] before aborting an ORTE operation (default: 1sec)",
false, false, 1, &value);
orte_max_timeout = 1000000.0 * value; /* convert to usec */
mca_base_param_reg_int_name("orte", "timeout_step",
"Time to wait [in usecs/proc] before aborting an ORTE operation (default: 1000 usec/proc)",
false, false, 1000, &orte_timeout_usec_per_proc);
#endif
/* default hostfile */
orte_default_hostfile = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "default_hostfile",
"Name of the default hostfile (relative or absolute path, \"none\" to ignore environmental or default MCA param setting)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_default_hostfile);
if (NULL == orte_default_hostfile) {
/* nothing was given, so define the default */
asprintf(&orte_default_hostfile, "%s/openmpi-default-hostfile", opal_install_dirs.sysconfdir);
/* flag that nothing was given */
orte_default_hostfile_given = false;
} else if (0 == strcmp(orte_default_hostfile, "none")) {
free (orte_default_hostfile);
orte_default_hostfile = NULL;
/* flag that it was given */
orte_default_hostfile_given = true;
} else {
/* flag that it was given */
orte_default_hostfile_given = true;
}
/* default dash-host */
orte_default_dash_host = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "default_dash_host",
"Default -host setting (specify \"none\" to ignore environmental or default MCA param setting)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_default_dash_host);
if (NULL != orte_default_dash_host &&
0 == strcmp(orte_default_dash_host, "none")) {
free(orte_default_dash_host);
orte_default_dash_host = NULL;
}
/* regex of nodes in system */
orte_node_regex = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "node_regex",
"Regular expression defining nodes in the system",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_node_regex);
/* whether or not to keep FQDN hostnames */
orte_keep_fqdn_hostnames = false;
(void) mca_base_var_register ("orte", "orte", NULL, "keep_fqdn_hostnames",
"Whether or not to keep FQDN hostnames [default: no]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_keep_fqdn_hostnames);
/* whether or not to retain aliases of hostnames */
orte_retain_aliases = false;
(void) mca_base_var_register ("orte", "orte", NULL, "retain_aliases",
"Whether or not to keep aliases for host names [default: no]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_retain_aliases);
/* which alias to use in MPIR_proctab */
orte_use_hostname_alias = 1;
(void) mca_base_var_register ("orte", "orte", NULL, "hostname_alias_index",
"If hostname aliases are being retained, which one to use for the debugger proc table [default: 1st alias]",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_use_hostname_alias);
orte_xml_output = false;
(void) mca_base_var_register ("orte", "orte", NULL, "xml_output",
"Display all output in XML format (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_xml_output);
/* whether to tag output */
/* if we requested xml output, be sure to tag the output as well */
orte_tag_output = orte_xml_output;
(void) mca_base_var_register ("orte", "orte", NULL, "tag_output",
"Tag all output with [job,rank] (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_tag_output);
if (orte_xml_output) {
orte_tag_output = true;
}
orte_xml_file = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "xml_file",
"Provide all output in XML format to the specified file",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_xml_file);
if (NULL != orte_xml_file) {
if (ORTE_PROC_IS_HNP && NULL == orte_xml_fp) {
/* only the HNP opens this file! Make sure it only happens once */
orte_xml_fp = fopen(orte_xml_file, "w");
if (NULL == orte_xml_fp) {
opal_output(0, "Could not open specified xml output file: %s", orte_xml_file);
return ORTE_ERROR;
}
}
/* ensure we set the flags to tag output */
orte_xml_output = true;
orte_tag_output = true;
} else {
/* default to stdout */
orte_xml_fp = stdout;
}
/* whether to timestamp output */
orte_timestamp_output = false;
(void) mca_base_var_register ("orte", "orte", NULL, "timestamp_output",
"Timestamp all application process output (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_timestamp_output);
/* redirect output into files */
orte_output_filename = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "output_filename",
"Redirect output from application processes into filename.rank [default: NULL]",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_output_filename);
orte_show_resolved_nodenames = false;
(void) mca_base_var_register ("orte", "orte", NULL, "show_resolved_nodenames",
"Display any node names that are resolved to a different name (default: false)",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_show_resolved_nodenames);
#if 0
/* XXX -- option doesn't appear to do anything */
mca_base_param_reg_int_name("orte", "hetero_apps",
"Indicates that multiple app_contexts are being provided that are a mix of 32/64 bit binaries (default: false)",
false, false, (int) false, &value);
orte_hetero_apps = OPAL_INT_TO_BOOL(value);
#endif
orte_hetero_nodes = false;
(void) mca_base_var_register ("orte", "orte", NULL, "hetero_nodes",
"Nodes in cluster may differ in topology, so send the topology back from each node [Default = false]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_hetero_nodes);
/* allow specification of the launch agent */
orte_launch_agent = "orted";
(void) mca_base_var_register ("orte", "orte", NULL, "launch_agent",
"Command used to start processes on remote nodes (default: orted)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_launch_agent);
orte_fork_agent_string = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "fork_agent",
"Command used to fork processes on remote nodes (default: NULL)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_fork_agent_string);
if (NULL != orte_fork_agent_string) {
orte_fork_agent = opal_argv_split(orte_fork_agent_string, ' ');
}
/* whether or not to require RM allocation */
orte_allocation_required = false;
(void) mca_base_var_register ("orte", "orte", NULL, "allocation_required",
"Whether or not an allocation by a resource manager is required [default: no]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_allocation_required);
/* whether or not to map stddiag to stderr */
orte_map_stddiag_to_stderr = false;
(void) mca_base_var_register ("orte", "orte", NULL, "map_stddiag_to_stderr",
"Map output from opal_output to stderr of the local process [default: no]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_map_stddiag_to_stderr);
/* generate new terminal windows to display output from specified ranks */
orte_xterm = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "xterm",
"Create a new xterm window and display output from the specified ranks there [default: none]",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_xterm);
if (NULL != orte_xterm) {
/* if an xterm request is given, we have to leave any ssh
* sessions attached so the xterm window manager can get
* back to the controlling terminal
*/
orte_leave_session_attached = true;
/* also want to redirect stddiag output from opal_output
* to stderr from the process so those messages show
* up in the xterm window instead of being forwarded to mpirun
*/
orte_map_stddiag_to_stderr = true;
}
/* whether or not to report launch progress */
orte_report_launch_progress = false;
(void) mca_base_var_register ("orte", "orte", NULL, "report_launch_progress",
"Output a brief periodic report on launch progress [default: no]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_report_launch_progress);
/* cluster hardware info detected by orte only */
orte_local_cpu_type = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "cpu_type",
"cpu type detected in node",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, MCA_BASE_VAR_FLAG_INTERNAL,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_local_cpu_type);
orte_local_cpu_model = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "cpu_model",
"cpu model detected in node",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, MCA_BASE_VAR_FLAG_INTERNAL,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_local_cpu_model);
/* tool communication controls */
orte_report_events_uri = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "report_events",
"URI to which events are to be reported (default: NULL)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_report_events_uri);
if (NULL != orte_report_events_uri) {
orte_report_events = true;
}
/* barrier control */
orte_do_not_barrier = false;
(void) mca_base_var_register ("orte", "orte", NULL, "do_not_barrier",
"Do not barrier in orte_init",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, MCA_BASE_VAR_FLAG_INTERNAL,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_do_not_barrier);
orte_enable_recovery = false;
(void) mca_base_var_register ("orte", "orte", NULL, "enable_recovery",
"Enable recovery from process failure [Default = disabled]",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_enable_recovery);
orte_max_restarts = 0;
(void) mca_base_var_register ("orte", "orte", NULL, "max_restarts",
"Max number of times to restart a failed process",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_max_restarts);
if (!orte_enable_recovery && orte_max_restarts != 0) {
if (ORTE_PROC_IS_HNP) {
opal_output(orte_clean_output,
"------------------------------------------------------------------\n"
"The MCA param orte_enable_recovery was not set to true, but\n"
"a value was provided for the number of restarts:\n\n"
"Max restarts: %d\n"
"We are enabling process recovery and continuing execution. To avoid\n"
"this warning in the future, please set the orte_enable_recovery\n"
"param to non-zero.\n"
"------------------------------------------------------------------",
orte_max_restarts);
}
orte_enable_recovery = true;
}
orte_abort_non_zero_exit = true;
(void) mca_base_var_register ("orte", "orte", NULL, "abort_on_non_zero_status",
"Abort the job if any process returns a non-zero exit status - no restart in such cases",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_abort_non_zero_exit);
orte_allowed_exit_without_sync = false;
(void) mca_base_var_register ("orte", "orte", NULL, "allowed_exit_without_sync",
"Process exiting without calling finalize will not trigger job termination",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_allowed_exit_without_sync);
orte_staged_execution = false;
(void) mca_base_var_register ("orte", "orte", NULL, "staged_execution",
"Staged execution is being used",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_staged_execution);
orte_report_child_jobs_separately = false;
(void) mca_base_var_register ("orte", "orte", NULL, "report_child_jobs_separately",
"Return the exit status of the primary job only",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_report_child_jobs_separately);
#if 0
/* XXX -- unused parameter */
mca_base_param_reg_int_name("orte", "child_time_to_exit",
"Max time a spawned child job is allowed to run after the primary job has terminated (seconds)",
false, false,
INT_MAX, &value);
orte_child_time_to_exit.tv_sec = value;
orte_child_time_to_exit.tv_usec = 0;
#endif
orte_stat_history_size = 1;
(void) mca_base_var_register ("orte", "orte", NULL, "stat_history_size",
"Number of stat samples to keep",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_stat_history_size);
orte_max_vm_size = -1;
(void) mca_base_var_register ("orte", "orte", NULL, "max_vm_size",
"Maximum size of virtual machine - used to subdivide allocation",
MCA_BASE_VAR_TYPE_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_max_vm_size);
if (opal_hwloc_use_hwthreads_as_cpus) {
orte_set_slots = "hwthreads";
} else {
orte_set_slots = "cores";
}
(void) mca_base_var_register ("orte", "orte", NULL, "set_default_slots",
"Set the number of slots on nodes that lack such info to the"
" number of specified objects [a number, \"cores\" (default),"
" \"numas\", \"sockets\", \"hwthreads\" (default if hwthreads_as_cpus is set),"
" or \"none\" to skip this option]",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_set_slots);
/* should we display the allocation after determining it? */
orte_display_allocation = false;
id = mca_base_var_register ("orte", "orte", NULL, "display_alloc",
"Whether to display the allocation after it is determined",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_display_allocation);
/* register a synonym for old name -- should we remove this now? */
mca_base_var_register_synonym (id, "orte", "ras", "base", "display_alloc", MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
/* should we display a detailed (developer-quality) version of the allocation after determining it? */
orte_devel_level_output = false;
id = mca_base_var_register ("orte", "orte", NULL, "display_devel_alloc",
"Whether to display a developer-detail allocation after it is determined",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_devel_level_output);
/* register a synonym for old name -- should we remove this now? */
mca_base_var_register_synonym (id, "orte", "ras", "base", "display_devel_alloc", MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
if (orte_devel_level_output) {
orte_display_allocation = true;
}
/* should we treat any -host directives as "soft" - i.e., desired
* but not required
*/
orte_soft_locations = false;
(void) mca_base_var_register ("orte", "orte", NULL, "soft_locations",
"Treat -host directives as desired, but not required",
MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_soft_locations);
/* allow specification of the cores to be used by daemons */
orte_daemon_cores = NULL;
(void) mca_base_var_register ("orte", "orte", NULL, "daemon_cores",
"Restrict the ORTE daemons (including mpirun) to operate on the specified cores (comma-separated list of ranges)",
MCA_BASE_VAR_TYPE_STRING, NULL, 0, 0,
OPAL_INFO_LVL_5, MCA_BASE_VAR_SCOPE_READONLY,
&orte_daemon_cores);
/* cutoff for full modex */
orte_direct_modex_cutoff = UINT32_MAX;
id = mca_base_var_register ("orte", "orte", NULL, "direct_modex_cutoff",
"If the number of processes in the application exceeds the provided value,"
"modex will be done upon demand [default: UINT32_MAX]",
MCA_BASE_VAR_TYPE_UNSIGNED_INT, NULL, 0, 0,
OPAL_INFO_LVL_9, MCA_BASE_VAR_SCOPE_READONLY,
&orte_direct_modex_cutoff);
/* register a synonym for old name */
mca_base_var_register_synonym (id, "ompi", "ompi", "hostname", "cutoff", MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
return ORTE_SUCCESS;
}
*
* The values below are the default values.
*/
bool ompi_mpi_param_check = true;
bool ompi_debug_show_handle_leaks = false;
int ompi_debug_show_mpi_alloc_mem_leaks = 0;
bool ompi_debug_no_free_handles = false;
bool ompi_mpi_show_mca_params = false;
char *ompi_mpi_show_mca_params_file = NULL;
bool ompi_mpi_abort_print_stack = false;
int ompi_mpi_abort_delay = 0;
bool ompi_mpi_keep_peer_hostnames = true;
bool ompi_mpi_keep_fqdn_hostnames = false;
int ompi_mpi_leave_pinned = -1;
bool ompi_mpi_leave_pinned_pipeline = false;
bool ompi_have_sparse_group_storage = OPAL_INT_TO_BOOL(OMPI_GROUP_SPARSE);
bool ompi_use_sparse_group_storage = OPAL_INT_TO_BOOL(OMPI_GROUP_SPARSE);
static bool show_default_mca_params = false;
static bool show_file_mca_params = false;
static bool show_enviro_mca_params = false;
static bool show_override_mca_params = false;
int ompi_mpi_register_params(void)
{
int value;
char *param;
/* Whether we want MPI API function parameter checking or not */
mca_base_param_reg_int_name("mpi", "param_check",