static void barrier_release(int status, orte_process_name_t* sender,
opal_buffer_t* buffer, orte_rml_tag_t tag,
void* cbdata)
{
int32_t cnt;
int rc, ret;
orte_grpcomm_signature_t *sig;
orte_grpcomm_coll_t *coll;
OPAL_OUTPUT_VERBOSE((5, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct: barrier release called with %d bytes",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), (int)buffer->bytes_used));
/* unpack the signature */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &sig, &cnt, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
return;
}
/* unpack the return status */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &ret, &cnt, OPAL_INT))) {
ORTE_ERROR_LOG(rc);
return;
}
/* check for the tracker - it is not an error if not
* found as that just means we wre not involved
* in the collective */
if (NULL == (coll = orte_grpcomm_base_get_tracker(sig, false))) {
OBJ_RELEASE(sig);
return;
}
/* execute the callback */
if (NULL != coll->cbfunc) {
coll->cbfunc(ret, buffer, coll->cbdata);
}
opal_list_remove_item(&orte_grpcomm_base.ongoing, &coll->super);
OBJ_RELEASE(coll);
OBJ_RELEASE(sig);
}
static void brks_allgather_recv_dist(int status, orte_process_name_t* sender,
opal_buffer_t* buffer, orte_rml_tag_t tag,
void* cbdata)
{
int32_t cnt, num_remote;
int rc;
orte_grpcomm_signature_t *sig;
orte_grpcomm_coll_t *coll;
orte_vpid_t distance, new_distance;
OPAL_OUTPUT_VERBOSE((5, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:coll:recdub received data",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
/* unpack the signature */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &sig, &cnt, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
return;
}
/* check for the tracker and create it if not found */
if (NULL == (coll = orte_grpcomm_base_get_tracker(sig, true))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
OBJ_RELEASE(sig);
return;
}
/* unpack the distance */
distance = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &distance, &cnt, OPAL_INT32))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
/* unpack number of reported processes */
num_remote = 0;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &num_remote, &cnt, OPAL_INT32))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
coll->nreported += num_remote;
/* capture any provided content */
if (OPAL_SUCCESS != (rc = opal_dss.copy_payload(&coll->bucket, buffer))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
//update distance and send
new_distance = distance <<= 1;
if (new_distance < coll->ndmns) {
brks_allgather_send_dist(coll, new_distance);
} else {
brks_finalize_coll(coll, ORTE_SUCCESS);
}
OBJ_RELEASE(sig);
return;
}
static void allgather_recv(int status, orte_process_name_t* sender,
opal_buffer_t* buffer, orte_rml_tag_t tag,
void* cbdata)
{
int32_t cnt;
int rc, ret;
orte_grpcomm_signature_t *sig;
opal_buffer_t *reply;
orte_grpcomm_coll_t *coll;
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather recvd from %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender)));
/* unpack the signature */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &sig, &cnt, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
return;
}
/* check for the tracker and create it if not found */
if (NULL == (coll = orte_grpcomm_base_get_tracker(sig, true))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
OBJ_RELEASE(sig);
return;
}
/* increment nprocs reported for collective */
coll->nreported++;
/* capture any provided content */
opal_dss.copy_payload(&coll->bucket, buffer);
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather recv nexpected %d nrep %d",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
(int)coll->nexpected, (int)coll->nreported));
/* see if everyone has reported */
if (coll->nreported == coll->nexpected) {
if (ORTE_PROC_IS_HNP) {
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather HNP reports complete",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
/* the allgather is complete - send the xcast */
reply = OBJ_NEW(opal_buffer_t);
/* pack the signature */
if (OPAL_SUCCESS != (rc = opal_dss.pack(reply, &sig, 1, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(reply);
OBJ_RELEASE(sig);
return;
}
/* pack the status - success since the allgather completed. This
* would be an error if we timeout instead */
ret = ORTE_SUCCESS;
if (OPAL_SUCCESS != (rc = opal_dss.pack(reply, &ret, 1, OPAL_INT))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(reply);
OBJ_RELEASE(sig);
return;
}
/* transfer the collected bucket */
opal_dss.copy_payload(reply, &coll->bucket);
/* send the release via xcast */
(void)orte_grpcomm.xcast(sig, ORTE_RML_TAG_COLL_RELEASE, reply);
OBJ_RELEASE(reply);
} else {
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather rollup complete - sending to %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(ORTE_PROC_MY_PARENT)));
/* relay the bucket upward */
reply = OBJ_NEW(opal_buffer_t);
/* pack the signature */
if (OPAL_SUCCESS != (rc = opal_dss.pack(reply, &sig, 1, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(reply);
OBJ_RELEASE(sig);
return;
}
/* transfer the collected bucket */
opal_dss.copy_payload(reply, &coll->bucket);
/* send the info to our parent */
rc = orte_rml.send_buffer_nb(ORTE_PROC_MY_PARENT, reply,
ORTE_RML_TAG_ALLGATHER_DIRECT,
orte_rml_send_callback, NULL);
}
}
OBJ_RELEASE(sig);
}
static void brks_allgather_recv_dist(int status, orte_process_name_t* sender,
opal_buffer_t* buffer, orte_rml_tag_t tag,
void* cbdata)
{
int32_t cnt;
int rc;
orte_grpcomm_signature_t *sig;
orte_grpcomm_coll_t *coll;
uint32_t distance;
OPAL_OUTPUT_VERBOSE((5, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:coll:brks RECEIVING FROM %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender)));
/* unpack the signature */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &sig, &cnt, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
return;
}
/* check for the tracker and create it if not found */
if (NULL == (coll = orte_grpcomm_base_get_tracker(sig, true))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
OBJ_RELEASE(sig);
return;
}
/* unpack the distance */
distance = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &distance, &cnt, OPAL_INT32))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
assert(0 == orte_grpcomm_base_check_distance_recv(coll, distance));
/* Check whether we can process next distance */
if (coll->nreported && (!distance || orte_grpcomm_base_check_distance_recv(coll, distance - 1))) {
size_t nreceived;
OPAL_OUTPUT_VERBOSE((80, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:coll:brks data from %d distance received, "
"Process the next distance.",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), distance));
/* capture any provided content */
rc = opal_dss.unpack (buffer, &nreceived, &cnt, OPAL_SIZE);
if (OPAL_SUCCESS != rc) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
if (OPAL_SUCCESS != (rc = opal_dss.copy_payload(&coll->bucket, buffer))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
coll->nreported += nreceived;
orte_grpcomm_base_mark_distance_recv(coll, distance);
brks_allgather_process_data(coll, distance + 1);
} else {
OPAL_OUTPUT_VERBOSE((80, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:coll:brks data from %d distance received, "
"still waiting for data.",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), distance));
if (NULL == coll->buffers) {
if (NULL == (coll->buffers = (opal_buffer_t **) calloc ((uint32_t) log2 (coll->ndmns) + 1, sizeof(opal_buffer_t *)))) {
rc = OPAL_ERR_OUT_OF_RESOURCE;
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
}
if (NULL == (coll->buffers[distance] = OBJ_NEW(opal_buffer_t))) {
rc = OPAL_ERR_OUT_OF_RESOURCE;
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
if (OPAL_SUCCESS != (rc = opal_dss.copy_payload(coll->buffers[distance], buffer))) {
OBJ_RELEASE(sig);
ORTE_ERROR_LOG(rc);
brks_finalize_coll(coll, rc);
return;
}
}
OBJ_RELEASE(sig);
}
static void allgather_recv(int status, orte_process_name_t* sender,
opal_buffer_t* buffer, orte_rml_tag_t tag,
void* cbdata)
{
int32_t cnt;
int rc, ret;
orte_grpcomm_signature_t *sig;
opal_buffer_t *reply;
orte_grpcomm_coll_t *coll;
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather recvd from %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender)));
/* unpack the signature */
cnt = 1;
if (OPAL_SUCCESS != (rc = opal_dss.unpack(buffer, &sig, &cnt, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
return;
}
/* check for the tracker and create it if not found */
if (NULL == (coll = orte_grpcomm_base_get_tracker(sig, true))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
OBJ_RELEASE(sig);
return;
}
/* increment nprocs reported for collective */
coll->nreported++;
/* capture any provided content */
opal_dss.copy_payload(&coll->bucket, buffer);
OPAL_OUTPUT_VERBOSE((1, orte_grpcomm_base_framework.framework_output,
"%s grpcomm:direct allgather recv ndmns %d nrep %d",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
(int)coll->ndmns, (int)coll->nreported));
/* if all participating daemons have reported */
if (coll->ndmns == coll->nreported) {
reply = OBJ_NEW(opal_buffer_t);
/* pack the signature */
if (OPAL_SUCCESS != (rc = opal_dss.pack(reply, &sig, 1, ORTE_SIGNATURE))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(reply);
OBJ_RELEASE(sig);
return;
}
/* pack the status - success since the allgather completed. This
* would be an error if we timeout instead */
ret = ORTE_SUCCESS;
if (OPAL_SUCCESS != (rc = opal_dss.pack(reply, &ret, 1, OPAL_INT))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(reply);
OBJ_RELEASE(sig);
return;
}
/* transfer the collected bucket */
opal_dss.copy_payload(reply, &coll->bucket);
/* send the release via xcast */
(void)orte_grpcomm.xcast(sig, ORTE_RML_TAG_COLL_RELEASE, reply);
OBJ_RELEASE(reply);
}
OBJ_RELEASE(sig);
}
