1
1
Gilles Gouaillardet 1c4a3598d0 pmix/pmix4x: refresh to the latest open PMIx master
refresh to openpmix/openpmix@ea3b29b1a4

Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
2019-10-01 14:27:22 +09:00

313 строки
12 KiB
C

/*
* Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2011 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
* Copyright (c) 2006-2013 Los Alamos National Security, LLC.
* All rights reserved.
* Copyright (c) 2009-2012 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2011 Oak Ridge National Labs. All rights reserved.
* Copyright (c) 2013-2018 Intel, Inc. All rights reserved.
* Copyright (c) 2015 Mellanox Technologies, Inc. All rights reserved.
* Copyright (c) 2019 Triad National Security, LLC. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*
*/
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <time.h>
#include <pthread.h>
#include <pmix.h>
typedef struct {
pthread_mutex_t mutex;
pthread_cond_t cond;
volatile bool active;
pmix_status_t status;
} mylock_t;
#define DEBUG_CONSTRUCT_LOCK(l) \
do { \
pthread_mutex_init(&(l)->mutex, NULL); \
pthread_cond_init(&(l)->cond, NULL); \
(l)->active = true; \
(l)->status = PMIX_SUCCESS; \
} while(0)
#define DEBUG_DESTRUCT_LOCK(l) \
do { \
pthread_mutex_destroy(&(l)->mutex); \
pthread_cond_destroy(&(l)->cond); \
} while(0)
#define DEBUG_WAIT_THREAD(lck) \
do { \
pthread_mutex_lock(&(lck)->mutex); \
while ((lck)->active) { \
pthread_cond_wait(&(lck)->cond, &(lck)->mutex); \
} \
pthread_mutex_unlock(&(lck)->mutex); \
} while(0)
#define DEBUG_WAKEUP_THREAD(lck) \
do { \
pthread_mutex_lock(&(lck)->mutex); \
(lck)->active = false; \
pthread_cond_broadcast(&(lck)->cond); \
pthread_mutex_unlock(&(lck)->mutex); \
} while(0)
static pmix_proc_t myproc;
static mylock_t invitedlock;
static void notification_fn(size_t evhdlr_registration_id,
pmix_status_t status,
const pmix_proc_t *source,
pmix_info_t info[], size_t ninfo,
pmix_info_t results[], size_t nresults,
pmix_event_notification_cbfunc_fn_t cbfunc,
void *cbdata)
{
fprintf(stderr, "Client %s:%d NOTIFIED with status %d\n", myproc.nspace, myproc.rank, status);
if (NULL != cbfunc) {
cbfunc(PMIX_EVENT_ACTION_COMPLETE, NULL, 0, NULL, NULL, cbdata);
}
}
static void op_callbk(pmix_status_t status,
void *cbdata)
{
mylock_t *lock = (mylock_t*)cbdata;
lock->status = status;
DEBUG_WAKEUP_THREAD(lock);
}
static void errhandler_reg_callbk(pmix_status_t status,
size_t errhandler_ref,
void *cbdata)
{
mylock_t *lock = (mylock_t*)cbdata;
lock->status = status;
DEBUG_WAKEUP_THREAD(lock);
}
static void grpcomplete(pmix_status_t status,
pmix_info_t *info, size_t ninfo,
void *cbdata,
pmix_release_cbfunc_t release_fn,
void *release_cbdata)
{
fprintf(stderr, "%s:%d GRPCOMPLETE\n", myproc.nspace, myproc.rank);
DEBUG_WAKEUP_THREAD(&invitedlock);
}
static void invitefn(size_t evhdlr_registration_id,
pmix_status_t status,
const pmix_proc_t *source,
pmix_info_t info[], size_t ninfo,
pmix_info_t results[], size_t nresults,
pmix_event_notification_cbfunc_fn_t cbfunc,
void *cbdata)
{
size_t n;
char *grp = NULL;
pmix_status_t rc;
/* if I am the leader, I can ignore this event */
if (PMIX_CHECK_PROCID(source, &myproc)) {
fprintf(stderr, "%s:%d INVITED, BUT LEADER\n", myproc.nspace, myproc.rank);
/* mark the event chain as complete */
if (NULL != cbfunc) {
cbfunc(PMIX_EVENT_ACTION_COMPLETE, NULL, 0, NULL, NULL, cbdata);
}
return;
}
/* search for grp id */
for (n=0; n < ninfo; n++) {
if (PMIX_CHECK_KEY(&info[n], PMIX_GROUP_ID)) {
grp = info[n].value.data.string;
break;
}
}
fprintf(stderr, "Client %s:%d INVITED by source %s:%d\n",
myproc.nspace, myproc.rank,
source->nspace, source->rank);
invitedlock.status = status;
fprintf(stderr, "%s:%d ACCEPTING INVITE\n", myproc.nspace, myproc.rank);
rc = PMIx_Group_join_nb(grp, source, PMIX_GROUP_ACCEPT, NULL, 0, grpcomplete, NULL);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "%s:%d Error in Group_join_nb: %sn", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
}
/* mark the event chain as complete */
if (NULL != cbfunc) {
cbfunc(PMIX_EVENT_ACTION_COMPLETE, NULL, 0, NULL, NULL, cbdata);
}
}
int main(int argc, char **argv)
{
int rc;
pmix_value_t value;
pmix_value_t *val = &value;
pmix_proc_t proc, *procs;
uint32_t nprocs;
mylock_t lock;
pmix_status_t code;
pmix_info_t *results;
size_t nresults;
/* init us */
if (PMIX_SUCCESS != (rc = PMIx_Init(&myproc, NULL, 0))) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Init failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
exit(0);
}
fprintf(stderr, "[%d] Client ns %s rank %d: Running\n", (int)getpid(), myproc.nspace, myproc.rank);
DEBUG_CONSTRUCT_LOCK(&invitedlock);
PMIX_PROC_CONSTRUCT(&proc);
(void)strncpy(proc.nspace, myproc.nspace, PMIX_MAX_NSLEN);
proc.rank = PMIX_RANK_WILDCARD;
/* get our universe size */
if (PMIX_SUCCESS != (rc = PMIx_Get(&proc, PMIX_UNIV_SIZE, NULL, 0, &val))) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Get universe size failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
goto done;
}
nprocs = val->data.uint32;
PMIX_VALUE_RELEASE(val);
if (nprocs < 4) {
if (0 == myproc.rank) {
fprintf(stderr, "This example requires a minimum of 4 processes\n");
}
goto done;
}
fprintf(stderr, "Client %s:%d universe size %d\n", myproc.nspace, myproc.rank, nprocs);
/* register our default errhandler */
DEBUG_CONSTRUCT_LOCK(&lock);
PMIx_Register_event_handler(NULL, 0, NULL, 0,
notification_fn, errhandler_reg_callbk, (void*)&lock);
DEBUG_WAIT_THREAD(&lock);
rc = lock.status;
DEBUG_DESTRUCT_LOCK(&lock);
if (PMIX_SUCCESS != rc) {
goto done;
}
/* we need to register handlers for invitations */
DEBUG_CONSTRUCT_LOCK(&lock);
code = PMIX_GROUP_INVITED;
PMIx_Register_event_handler(&code, 1, NULL, 0,
invitefn, errhandler_reg_callbk, (void*)&lock);
DEBUG_WAIT_THREAD(&lock);
rc = lock.status;
DEBUG_DESTRUCT_LOCK(&lock);
if (PMIX_SUCCESS != rc) {
goto done;
}
/* call fence to sync */
PMIX_PROC_CONSTRUCT(&proc);
(void)strncpy(proc.nspace, myproc.nspace, PMIX_MAX_NSLEN);
proc.rank = PMIX_RANK_WILDCARD;
if (PMIX_SUCCESS != (rc = PMIx_Fence(&proc, 1, NULL, 0))) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Fence failed: %d\n", myproc.nspace, myproc.rank, rc);
goto done;
}
/* rank=0 constructs a new group */
if (0 == myproc.rank) {
fprintf(stderr, "%d executing Group_invite\n", myproc.rank);
nprocs = 3;
PMIX_PROC_CREATE(procs, nprocs);
PMIX_PROC_LOAD(&procs[0], myproc.nspace, 0);
PMIX_PROC_LOAD(&procs[1], myproc.nspace, 2);
PMIX_PROC_LOAD(&procs[2], myproc.nspace, 3);
rc = PMIx_Group_invite("ourgroup", procs, nprocs, NULL, 0, &results, &nresults);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Group_invite failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
goto done;
}
PMIX_PROC_FREE(procs, nprocs);
fprintf(stderr, "%s:%d Execute fence across group\n", myproc.nspace, myproc.rank);
PMIX_PROC_LOAD(&proc, "ourgroup", PMIX_RANK_WILDCARD);
rc = PMIx_Fence(&proc, 1, NULL, 0);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Fence across group failed: %d\n", myproc.nspace, myproc.rank, rc);
goto done;
}
fprintf(stderr, "%d executing Group_destruct\n", myproc.rank);
rc = PMIx_Group_destruct("ourgroup", NULL, 0);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Group_destruct failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
goto done;
}
} else if (2 == myproc.rank || 3 == myproc.rank) {
/* wait to be invited */
fprintf(stderr, "%s:%d waiting to be invited\n", myproc.nspace, myproc.rank);
DEBUG_WAIT_THREAD(&invitedlock);
DEBUG_DESTRUCT_LOCK(&invitedlock);
fprintf(stderr, "%s:%d Execute fence across group\n", myproc.nspace, myproc.rank);
PMIX_PROC_LOAD(&proc, "ourgroup", PMIX_RANK_WILDCARD);
rc = PMIx_Fence(&proc, 1, NULL, 0);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Fence across group failed: %d\n", myproc.nspace, myproc.rank, rc);
goto done;
}
fprintf(stderr, "%d executing Group_destruct\n", myproc.rank);
rc = PMIx_Group_destruct("ourgroup", NULL, 0);
if (PMIX_SUCCESS != rc) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Group_destruct failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
goto done;
}
}
/* call fence to sync */
PMIX_PROC_CONSTRUCT(&proc);
(void)strncpy(proc.nspace, myproc.nspace, PMIX_MAX_NSLEN);
proc.rank = PMIX_RANK_WILDCARD;
if (PMIX_SUCCESS != (rc = PMIx_Fence(&proc, 1, NULL, 0))) {
fprintf(stderr, "Client ns %s rank %d: PMIx_Fence failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
goto done;
}
done:
/* finalize us */
DEBUG_CONSTRUCT_LOCK(&lock);
PMIx_Deregister_event_handler(1, op_callbk, &lock);
DEBUG_WAIT_THREAD(&lock);
DEBUG_DESTRUCT_LOCK(&lock);
fprintf(stderr, "Client ns %s rank %d: Finalizing\n", myproc.nspace, myproc.rank);
if (PMIX_SUCCESS != (rc = PMIx_Finalize(NULL, 0))) {
fprintf(stderr, "Client ns %s rank %d:PMIx_Finalize failed: %s\n", myproc.nspace, myproc.rank, PMIx_Error_string(rc));
} else {
fprintf(stderr, "Client ns %s rank %d:PMIx_Finalize successfully completed\n", myproc.nspace, myproc.rank);
}
fprintf(stderr, "%s:%d COMPLETE\n", myproc.nspace, myproc.rank);
fflush(stderr);
return(0);
}