static int local_setup(void)
{
int ret;
char *error = NULL;
int value;
orte_proc_t *proc;
orte_node_t *node;
orte_app_context_t *app;
/* initialize the global list of local children and job data */
OBJ_CONSTRUCT(&orte_local_children, opal_list_t);
OBJ_CONSTRUCT(&orte_local_jobdata, opal_list_t);
/* determine the topology info */
if (0 == orte_default_num_sockets_per_board) {
/* we weren't given a number, so try to determine it */
if (OPAL_SUCCESS != opal_paffinity_base_get_socket_info(&value)) {
/* can't get any info - default to 1 */
value = 1;
}
orte_default_num_sockets_per_board = (uint8_t)value;
}
if (0 == orte_default_num_cores_per_socket) {
/* we weren't given a number, so try to determine it */
if (OPAL_SUCCESS != opal_paffinity_base_get_core_info(0, &value)) {
/* don't have topo info - can we at least get #processors? */
if (OPAL_SUCCESS != opal_paffinity_base_get_processor_info(&value)) {
/* can't get any info - default to 1 */
value = 1;
}
}
orte_default_num_cores_per_socket = (uint8_t)value;
}
/* setup the global job and node arrays */
orte_job_data = OBJ_NEW(opal_pointer_array_t);
if (ORTE_SUCCESS != (ret = opal_pointer_array_init(orte_job_data,
1,
ORTE_GLOBAL_ARRAY_MAX_SIZE,
1))) {
ORTE_ERROR_LOG(ret);
error = "setup job array";
goto error;
}
orte_node_pool = OBJ_NEW(opal_pointer_array_t);
if (ORTE_SUCCESS != (ret = opal_pointer_array_init(orte_node_pool,
ORTE_GLOBAL_ARRAY_BLOCK_SIZE,
ORTE_GLOBAL_ARRAY_MAX_SIZE,
ORTE_GLOBAL_ARRAY_BLOCK_SIZE))) {
ORTE_ERROR_LOG(ret);
error = "setup node array";
goto error;
}
/* Setup the job data object for the daemons */
/* create and store the job data object */
daemons = OBJ_NEW(orte_job_t);
daemons->jobid = ORTE_PROC_MY_NAME->jobid;
daemons->name = strdup("ORCM DVM");
daemons->instance = strdup(ORTE_JOBID_PRINT(ORTE_PROC_MY_NAME->jobid));
/* create an app */
app = OBJ_NEW(orte_app_context_t);
app->app = strdup("orcmd");
opal_argv_append_nosize(&app->argv, "orcmd");
/* add to the daemon job - always must be an app for a job */
opal_pointer_array_add(daemons->apps, app);
/* setup the daemon map so it knows how to map them */
daemons->map = OBJ_NEW(orte_job_map_t);
daemons->map->policy = ORTE_MAPPING_BYNODE;
/* save it */
opal_pointer_array_set_item(orte_job_data, 0, daemons);
/* ensure our mapping policy will utilize any VM */
ORTE_ADD_MAPPING_POLICY(ORTE_MAPPING_USE_VM);
/* use bynode mapping by default */
ORTE_ADD_MAPPING_POLICY(ORTE_MAPPING_BYNODE);
/* create and store a node object where we are */
node = OBJ_NEW(orte_node_t);
node->name = strdup(orte_process_info.nodename);
node->slots = 1; /* min number */
node->slots_alloc = node->slots;
node->index = ORTE_PROC_MY_NAME->vpid;
opal_pointer_array_set_item(orte_node_pool, ORTE_PROC_MY_NAME->vpid, node);
/* create and store a proc object for us */
proc = OBJ_NEW(orte_proc_t);
proc->name.jobid = ORTE_PROC_MY_NAME->jobid;
proc->name.vpid = ORTE_PROC_MY_NAME->vpid;
proc->pid = orte_process_info.pid;
proc->state = ORTE_PROC_STATE_RUNNING;
OBJ_RETAIN(node); /* keep accounting straight */
proc->node = node;
proc->nodename = node->name;
opal_pointer_array_set_item(daemons->procs, proc->name.vpid, proc);
/* record that the daemon (i.e., us) is on this node
* NOTE: we do not add the proc object to the node's
* proc array because we are not an application proc.
* Instead, we record it in the daemon field of the
* node object
*/
OBJ_RETAIN(proc); /* keep accounting straight */
node->daemon = proc;
node->daemon_launched = true;
node->state = ORTE_NODE_STATE_UP;
/* record that the daemon job is running */
daemons->num_procs = 1;
daemons->state = ORTE_JOB_STATE_RUNNING;
/* open and setup the opal_pstat framework so we can provide
* process stats if requested
*/
if (ORTE_SUCCESS != (ret = opal_pstat_base_open())) {
ORTE_ERROR_LOG(ret);
error = "opal_pstat_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = opal_pstat_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_pstat_base_select";
goto error;
}
/* open and setup the local resource discovery framework */
if (ORTE_SUCCESS != (ret = opal_sysinfo_base_open())) {
ORTE_ERROR_LOG(ret);
error = "opal_sysinfo_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = opal_sysinfo_base_select())) {
ORTE_ERROR_LOG(ret);
error = "opal_sysinfo_base_select";
goto error;
}
/* Setup the communication infrastructure */
/* Runtime Messaging Layer - this opens/selects the OOB as well */
if (ORTE_SUCCESS != (ret = orte_rml_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_rml_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml_base_select";
goto error;
}
/* Routed system */
if (ORTE_SUCCESS != (ret = orte_routed_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_routed_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_routed_base_select";
goto error;
}
/* multicast */
if (ORTE_SUCCESS != (ret = orte_rmcast_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_rmcast_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_rmcast_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_rmcast_base_select";
goto error;
}
/* Open/select the odls */
if (ORTE_SUCCESS != (ret = orte_odls_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_odls_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_odls_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_odls_base_select";
goto error;
}
/* enable communication with the rml */
if (ORTE_SUCCESS != (ret = orte_rml.enable_comm())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml.enable_comm";
goto error;
}
/* insert our contact info into our process_info struct so we
* have it for later use and set the local daemon field to our name
*/
orte_process_info.my_daemon_uri = orte_rml.get_contact_info();
ORTE_PROC_MY_DAEMON->jobid = ORTE_PROC_MY_NAME->jobid;
ORTE_PROC_MY_DAEMON->vpid = ORTE_PROC_MY_NAME->vpid;
proc->rml_uri = orte_rml.get_contact_info();
/* setup the pnp framework */
if (ORCM_SUCCESS != (ret = orcm_pnp_base_open())) {
error = "pnp_open";
goto error;
}
if (ORCM_SUCCESS != (ret = orcm_pnp_base_select())) {
error = "pnp_select";
goto error;
}
/* setup the leader framework */
if (ORCM_SUCCESS != (ret = orcm_leader_base_open())) {
error = "leader_open";
goto error;
}
if (ORCM_SUCCESS != (ret = orcm_leader_base_select())) {
error = "leader_select";
goto error;
}
/* set the communication function */
orte_comm = orte_global_comm;
/* update the routing tree */
if (ORTE_SUCCESS != (ret = orte_routed.update_routing_tree())) {
ORTE_ERROR_LOG(ret);
error = "failed to update routing tree";
goto error;
}
/* setup the routed info - the selected routed component
* will know what to do.
*/
if (ORTE_SUCCESS != (ret = orte_routed.init_routes(ORTE_PROC_MY_NAME->jobid, NULL))) {
ORTE_ERROR_LOG(ret);
error = "orte_routed.init_routes";
goto error;
}
/* open/select the errmgr - do this after the daemon job
* has been defined so that the errmgr can get that
* job object
*/
if (ORTE_SUCCESS != (ret = orte_errmgr_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_errmgr_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_errmgr_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_errmgr_base_select";
goto error;
}
/* We actually do *not* want an orcm to voluntarily yield() the
processor more than necessary. orcm already blocks when
it is doing nothing, so it doesn't use any more CPU cycles than
it should; but when it *is* doing something, we do not want it
to be unnecessarily delayed because it voluntarily yielded the
processor in the middle of its work.
For example: when a message arrives at orcm, we want the
OS to wake us up in a timely fashion (which most OS's
seem good about doing) and then we want orcm to process
the message as fast as possible. If orcm yields and lets
aggressive applications get the processor back, it may be a
long time before the OS schedules orcm to run again
(particularly if there is no IO event to wake it up). Hence,
routed OOB messages (for example) may be significantly delayed
before being delivered to processes, which can be
problematic in some scenarios */
opal_progress_set_yield_when_idle(false);
/* Change the default behavior of libevent such that we want to
continually block rather than blocking for the default timeout
and then looping around the progress engine again. There
should be nothing in the orcmd that cannot block in libevent
until "something" happens (i.e., there's no need to keep
cycling through progress because the only things that should
happen will happen in libevent). This is a minor optimization,
but what the heck... :-) */
opal_progress_set_event_flag(OPAL_EVLOOP_ONCE);
/* setup the primary daemon command receive function */
ret = orte_rml.recv_buffer_nb(ORTE_NAME_WILDCARD, ORTE_RML_TAG_DAEMON,
ORTE_RML_NON_PERSISTENT, orte_daemon_recv, NULL);
if (ret != ORTE_SUCCESS && ret != ORTE_ERR_NOT_IMPLEMENTED) {
ORTE_ERROR_LOG(ret);
error = "daemon_recv";
goto error;
}
/* output a message indicating we are alive, our name, and our pid
* for debugging purposes
*/
if (orte_debug_daemons_flag) {
fprintf(stderr, "%s checking in as pid %ld on host %s\n",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), (long)orte_process_info.pid,
orte_process_info.nodename);
}
/* listen for termination cmds */
if (ORCM_SUCCESS != (ret = orcm_pnp.register_receive("orcm", "0.1", "alpha",
ORCM_PNP_SYS_CHANNEL,
ORCM_PNP_TAG_TERMINATE,
vm_term, NULL))) {
error = "orcm recv";
goto error;
}
if (ORCM_SUCCESS != (ret = orcm_pnp.register_receive("orcm-stop", "0.1", "alpha",
ORCM_PNP_SYS_CHANNEL,
ORCM_PNP_TAG_TERMINATE,
vm_term, NULL))) {
error = "orcm-stop recv";
goto error;
}
/* register to catch vm commands requests */
if (ORCM_SUCCESS != (ret = orcm_pnp.register_receive("orcm-sched", "0.1", "alpha",
ORCM_PNP_SYS_CHANNEL,
ORCM_PNP_TAG_COMMAND,
vm_cmd, NULL))) {
error = "orcm-sched recv";
goto error;
}
/* listen for state data requests */
if (ORCM_SUCCESS != (ret = orcm_pnp.register_receive("orcm-sched", "0.1", "alpha",
ORCM_PNP_SYS_CHANNEL,
ORCM_PNP_TAG_BOOTSTRAP,
recv_contact, NULL))) {
error = "contact recv";
goto error;
}
/* setup the SENSOR framework */
if (ORTE_SUCCESS != (ret = orte_sensor_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_sensor_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_sensor_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_sensor_select";
goto error;
}
/* announce our existence - this carries with it our rml uri and
* our local node system info
*/
if (ORCM_SUCCESS != (ret = orcm_pnp.announce("ORCMD", "0.1", "alpha", vm_tracker))) {
ORTE_ERROR_LOG(ret);
error = "announce";
goto error;
}
return ORTE_SUCCESS;
error:
orte_show_help("help-orte-runtime.txt",
"orte_init:startup:internal-failure",
true, error, ORTE_ERROR_NAME(ret), ret);
return ret;
}
static int local_init(void)
{
int ret;
char *error;
/* Setup the communication infrastructure */
/* Runtime Messaging Layer */
if (ORTE_SUCCESS != (ret = orte_rml_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_rml_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml_base_select";
goto error;
}
/* Routed system */
if (ORTE_SUCCESS != (ret = orte_routed_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_routed_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_routed_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_routed_base_select";
goto error;
}
/* multicast */
if (ORTE_SUCCESS != (ret = orte_rmcast_base_open())) {
ORTE_ERROR_LOG(ret);
error = "orte_rmcast_base_open";
goto error;
}
if (ORTE_SUCCESS != (ret = orte_rmcast_base_select())) {
ORTE_ERROR_LOG(ret);
error = "orte_rmcast_base_select";
goto error;
}
/* setup the pnp framework */
if (ORCM_SUCCESS != (ret = orcm_pnp_base_open())) {
error = "pnp_open";
goto error;
}
if (ORCM_SUCCESS != (ret = orcm_pnp_base_select())) {
error = "pnp_select";
goto error;
}
/* setup the leader framework */
if (ORCM_SUCCESS != (ret = orcm_leader_base_open())) {
error = "leader_open";
goto error;
}
if (ORCM_SUCCESS != (ret = orcm_leader_base_select())) {
error = "leader_select";
goto error;
}
/* enable communication via the rml */
if (ORTE_SUCCESS != (ret = orte_rml.enable_comm())) {
ORTE_ERROR_LOG(ret);
error = "orte_rml.enable_comm";
goto error;
}
/* setup the routed info - the selected routed component
* will know what to do.
*/
if (ORTE_SUCCESS != (ret = orte_routed.init_routes(ORTE_PROC_MY_NAME->jobid, NULL))) {
ORTE_ERROR_LOG(ret);
error = "orte_routed.init_routes";
goto error;
}
return ORTE_SUCCESS;
error:
orte_show_help("help-orte-runtime.txt",
"orte_init:startup:internal-failure",
true, error, ORTE_ERROR_NAME(ret), ret);
return ret;
}
