-/* Copyright (c) 2010. The SimGrid Team.
+/* Copyright (c) 2010-2014. The SimGrid Team.
* All rights reserved. */
/* This program is free software; you can redistribute it and/or modify it
- * under the terms of the license (GNU LGPL) which comes with this package. */
+ * under the terms of the license (GNU LGPL) which comes with this package. */
#include <stdlib.h>
#include "private.h"
+#include "xbt/dict.h"
#include "smpi_mpi_dt_private.h"
+#include "limits.h"
+#include "simix/smx_private.h"
+#include "colls/colls.h"
+#include "xbt/ex.h"
XBT_LOG_NEW_DEFAULT_SUBCATEGORY(smpi_comm, smpi,
"Logging specific to SMPI (comm)");
+xbt_dict_t smpi_comm_keyvals = NULL;
+int comm_keyval_id = 0;//avoid collisions
+
+/* Support for cartesian topology was added, but there are 2 other types of
+ * topology, graph et dist graph. In order to support them, we have to add a
+ * field MPIR_Topo_type, and replace the MPI_Topology field by an union. */
+
typedef struct s_smpi_mpi_communicator {
MPI_Group group;
+ MPIR_Topo_type topoType;
+ MPI_Topology topo; // to be replaced by an union
+ int refcount;
+ MPI_Comm leaders_comm;//inter-node communicator
+ MPI_Comm intra_comm;//intra-node communicator . For MPI_COMM_WORLD this can't be used, as var is global.
+ //use an intracomm stored in the process data instead
+ int* leaders_map; //who is the leader of each process
+ int is_uniform;
+ int* non_uniform_map; //set if smp nodes have a different number of processes allocated
+ int is_blocked;// are ranks allocated on the same smp node contiguous ?
+ xbt_dict_t attributes;
} s_smpi_mpi_communicator_t;
static int smpi_compare_rankmap(const void *a, const void *b)
return 1;
}
-MPI_Comm smpi_comm_new(MPI_Group group)
+MPI_Comm smpi_comm_new(MPI_Group group, MPI_Topology topo)
{
MPI_Comm comm;
comm = xbt_new(s_smpi_mpi_communicator_t, 1);
comm->group = group;
smpi_group_use(comm->group);
+ comm->refcount=1;
+ comm->topoType = -1;
+ comm->topo = topo;
+ comm->intra_comm = MPI_COMM_NULL;
+ comm->leaders_comm = MPI_COMM_NULL;
+ comm->is_uniform=1;
+ comm->non_uniform_map = NULL;
+ comm->leaders_map = NULL;
+ comm->is_blocked=0;
+ comm->attributes=NULL;
return comm;
}
void smpi_comm_destroy(MPI_Comm comm)
{
- xbt_free(comm);
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ smpi_group_unuse(comm->group);
+ smpi_topo_destroy(comm->topo); // there's no use count on topos
+ smpi_comm_unuse(comm);
+}
+
+int smpi_comm_dup(MPI_Comm comm, MPI_Comm* newcomm){
+ if(smpi_privatize_global_variables){ //we need to switch here, as the called function may silently touch global variables
+ smpi_switch_data_segment(smpi_process_index());
+ }
+ (*newcomm) = smpi_comm_new(smpi_comm_group(comm), smpi_comm_topo(comm));
+ int ret = MPI_SUCCESS;
+ //todo: faire en sorte que ça fonctionne avec un communicator dupliqué (refaire un init_smp ?)
+
+ /* MPI_Comm tmp=smpi_comm_get_intra_comm(comm);
+ if( tmp != MPI_COMM_NULL)
+ smpi_comm_set_intra_comm((*newcomm), smpi_comm_dup(tmp));
+ tmp=smpi_comm_get_leaders_comm(comm);
+ if( tmp != MPI_COMM_NULL)
+ smpi_comm_set_leaders_comm((*newcomm), smpi_comm_dup(tmp));
+ if(comm->non_uniform_map !=NULL){
+ (*newcomm)->non_uniform_map=
+ xbt_malloc(smpi_comm_size(comm->leaders_comm)*sizeof(int));
+ memcpy((*newcomm)->non_uniform_map,
+ comm->non_uniform_map,smpi_comm_size(comm->leaders_comm)*sizeof(int) );
+ }
+ if(comm->leaders_map !=NULL){
+ (*newcomm)->leaders_map=xbt_malloc(smpi_comm_size(comm)*sizeof(int));
+ memcpy((*newcomm)->leaders_map,
+ comm->leaders_map,smpi_comm_size(comm)*sizeof(int) );
+ }*/
+ if(comm->attributes !=NULL){
+ (*newcomm)->attributes=xbt_dict_new();
+ xbt_dict_cursor_t cursor = NULL;
+ int *key;
+ int flag;
+ void* value_in;
+ void* value_out;
+ xbt_dict_foreach(comm->attributes, cursor, key, value_in){
+ smpi_comm_key_elem elem = xbt_dict_get_or_null_ext(smpi_comm_keyvals, (const char*)key, sizeof(int));
+ if(elem && elem->copy_fn!=MPI_NULL_COPY_FN){
+ ret = elem->copy_fn(comm, *key, NULL, value_in, &value_out, &flag );
+ if(ret!=MPI_SUCCESS){
+ smpi_comm_destroy(*newcomm);
+ *newcomm=MPI_COMM_NULL;
+ return ret;
+ }
+ if(flag)
+ xbt_dict_set_ext((*newcomm)->attributes, (const char*)key, sizeof(int),value_out, NULL);
+ }
+ }
+ }
+ return ret;
}
+
MPI_Group smpi_comm_group(MPI_Comm comm)
{
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+
return comm->group;
}
+MPI_Topology smpi_comm_topo(MPI_Comm comm) {
+ if (comm != MPI_COMM_NULL)
+ return comm->topo;
+ return NULL;
+}
+
int smpi_comm_size(MPI_Comm comm)
{
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+
return smpi_group_size(smpi_comm_group(comm));
}
int smpi_comm_rank(MPI_Comm comm)
{
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
return smpi_group_rank(smpi_comm_group(comm), smpi_process_index());
}
void smpi_comm_get_name (MPI_Comm comm, char* name, int* len)
{
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
if(comm == MPI_COMM_WORLD) {
strcpy(name, "WORLD");
*len = 5;
}
}
+void smpi_comm_set_leaders_comm(MPI_Comm comm, MPI_Comm leaders){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ comm->leaders_comm=leaders;
+}
+
+void smpi_comm_set_intra_comm(MPI_Comm comm, MPI_Comm leaders){
+ comm->intra_comm=leaders;
+}
+
+int* smpi_comm_get_non_uniform_map(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ return comm->non_uniform_map;
+}
+
+int* smpi_comm_get_leaders_map(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ return comm->leaders_map;
+}
+
+MPI_Comm smpi_comm_get_leaders_comm(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ return comm->leaders_comm;
+}
+
+MPI_Comm smpi_comm_get_intra_comm(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED || comm==MPI_COMM_WORLD)
+ return smpi_process_get_comm_intra();
+ else return comm->intra_comm;
+}
+
+int smpi_comm_is_uniform(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ return comm->is_uniform;
+}
+
+int smpi_comm_is_blocked(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ return comm->is_blocked;
+}
+
MPI_Comm smpi_comm_split(MPI_Comm comm, int color, int key)
{
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
int system_tag = 123;
int index, rank, size, i, j, count, reqs;
int* sendbuf;
group_root = group_out; /* Save root's group */
}
for(j = 0; j < count; j++) {
+ //increment refcounter in order to avoid freeing the group too quick before copy
index = smpi_group_index(group, rankmap[2 * j]);
smpi_group_set_mapping(group_out, index, j);
}
} else {
if(color != MPI_UNDEFINED) {
smpi_mpi_recv(&group_out, 1, MPI_PTR, 0, system_tag, comm, MPI_STATUS_IGNORE);
+ if(group_out){
+ group_out=smpi_group_copy(group_out);
+ }
} /* otherwise, exit with group_out == NULL */
}
- return group_out ? smpi_comm_new(group_out) : MPI_COMM_NULL;
+ return group_out ? smpi_comm_new(group_out, NULL) : MPI_COMM_NULL;
+}
+
+void smpi_comm_use(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ comm->refcount++;
+}
+
+void smpi_comm_unuse(MPI_Comm comm){
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+ comm->refcount--;
+ if(comm->refcount==0){
+ if(comm->intra_comm != MPI_COMM_NULL)
+ smpi_comm_unuse(comm->intra_comm);
+ if(comm->leaders_comm != MPI_COMM_NULL)
+ smpi_comm_unuse(comm->leaders_comm);
+ if(comm->non_uniform_map !=NULL)
+ xbt_free(comm->non_uniform_map);
+ if(comm->leaders_map !=NULL)
+ xbt_free(comm->leaders_map);
+ if(comm->attributes !=NULL){
+ xbt_dict_cursor_t cursor = NULL;
+ int* key;
+ void * value;
+ int flag;
+ xbt_dict_foreach(comm->attributes, cursor, key, value){
+ smpi_comm_key_elem elem = xbt_dict_get_or_null(smpi_comm_keyvals, (const char*)key);
+ if(elem && elem->delete_fn)
+ elem->delete_fn(comm, *key, value, &flag);
+ }
+ }
+ xbt_free(comm);
+ }
+}
+
+static int
+compare_ints (const void *a, const void *b)
+{
+ const int *da = (const int *) a;
+ const int *db = (const int *) b;
+
+ return (*da > *db) - (*da < *db);
+}
+
+void smpi_comm_init_smp(MPI_Comm comm){
+ int leader = -1;
+
+ if (comm == MPI_COMM_UNINITIALIZED)
+ comm = smpi_process_comm_world();
+
+ int comm_size =smpi_comm_size(comm);
+
+ // If we are in replay - perform an ugly hack
+ // say to SimGrid that we are not in replay for a while, because we need
+ // the buffers to be copied for the following calls
+ int replaying = 0; //cache data to set it back again after
+ if(smpi_process_get_replaying()){
+ replaying=1;
+ smpi_process_set_replaying(0);
+ }
+
+ if(smpi_privatize_global_variables){ //we need to switch here, as the called function may silently touch global variables
+ smpi_switch_data_segment(smpi_process_index());
+ }
+ //identify neighbours in comm
+ //get the indexes of all processes sharing the same simix host
+ xbt_swag_t process_list = simcall_host_get_process_list(SIMIX_host_self());
+ int intra_comm_size = 0;
+ //only one process/node, disable SMP support and return
+// if(intra_comm_size==1){
+// smpi_comm_set_intra_comm(comm, MPI_COMM_SELF);
+// //smpi_comm_set_leaders_comm(comm, comm);
+// smpi_process_set_comm_intra(MPI_COMM_SELF);
+// return;
+// }
+
+
+ int i =0;
+ int min_index=INT_MAX;//the minimum index will be the leader
+ smx_process_t process = NULL;
+ xbt_swag_foreach(process, process_list) {
+ //is_in_comm=0;
+ int index = SIMIX_process_get_PID(process) -1;
+
+ if(smpi_group_rank(smpi_comm_group(comm), index)!=MPI_UNDEFINED){
+ intra_comm_size++;
+ //the process is in the comm
+ if(index < min_index)
+ min_index=index;
+ i++;
+ }
+ }
+ XBT_DEBUG("number of processes deployed on my node : %d", intra_comm_size);
+ MPI_Group group_intra = smpi_group_new(intra_comm_size);
+ i=0;
+ process = NULL;
+ xbt_swag_foreach(process, process_list) {
+ //is_in_comm=0;
+ int index = SIMIX_process_get_PID(process) -1;
+ if(smpi_group_rank(smpi_comm_group(comm), index)!=MPI_UNDEFINED){
+ smpi_group_set_mapping(group_intra, index, i);
+ i++;
+ }
+ }
+
+
+ MPI_Comm comm_intra = smpi_comm_new(group_intra, NULL);
+ //MPI_Comm shmem_comm = smpi_process_comm_intra();
+ //int intra_rank = smpi_comm_rank(shmem_comm);
+
+
+ //if(smpi_process_index()==min_index)
+ leader=min_index;
+
+ int * leaders_map= (int*)xbt_malloc0(sizeof(int)*comm_size);
+ int * leader_list= (int*)xbt_malloc0(sizeof(int)*comm_size);
+ for(i=0; i<comm_size; i++){
+ leader_list[i]=-1;
+ }
+
+ smpi_coll_tuned_allgather_mpich(&leader, 1, MPI_INT , leaders_map, 1, MPI_INT, comm);
+
+ if(smpi_privatize_global_variables){ //we need to switch here, as the called function may silently touch global variables
+ smpi_switch_data_segment(smpi_process_index());
+ }
+
+ if(!comm->leaders_map){
+ comm->leaders_map= leaders_map;
+ }else{
+ xbt_free(leaders_map);
+ }
+ int j=0;
+ int leader_group_size = 0;
+ for(i=0; i<comm_size; i++){
+ int already_done=0;
+ for(j=0;j<leader_group_size; j++){
+ if(comm->leaders_map[i]==leader_list[j]){
+ already_done=1;
+ }
+ }
+ if(!already_done){
+ leader_list[leader_group_size]=comm->leaders_map[i];
+ leader_group_size++;
+ }
+ }
+ qsort(leader_list, leader_group_size, sizeof(int),compare_ints);
+
+ MPI_Group leaders_group = smpi_group_new(leader_group_size);
+
+
+ MPI_Comm leader_comm = MPI_COMM_NULL;
+ if(MPI_COMM_WORLD!=MPI_COMM_UNINITIALIZED && comm!=MPI_COMM_WORLD){
+ //create leader_communicator
+ for (i=0; i< leader_group_size;i++)
+ smpi_group_set_mapping(leaders_group, leader_list[i], i);
+ leader_comm = smpi_comm_new(leaders_group, NULL);
+ smpi_comm_set_leaders_comm(comm, leader_comm);
+ smpi_comm_set_intra_comm(comm, comm_intra);
+
+ //create intracommunicator
+ // smpi_comm_set_intra_comm(comm, smpi_comm_split(comm, *(int*)SIMIX_host_self(), comm_rank));
+ }else{
+ for (i=0; i< leader_group_size;i++)
+ smpi_group_set_mapping(leaders_group, leader_list[i], i);
+
+ leader_comm = smpi_comm_new(leaders_group, NULL);
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL)
+ smpi_comm_set_leaders_comm(comm, leader_comm);
+ smpi_process_set_comm_intra(comm_intra);
+ }
+
+ int is_uniform = 1;
+
+ // Are the nodes uniform ? = same number of process/node
+ int my_local_size=smpi_comm_size(comm_intra);
+ if(smpi_comm_rank(comm_intra)==0) {
+ int* non_uniform_map = xbt_malloc0(sizeof(int)*leader_group_size);
+ smpi_coll_tuned_allgather_mpich(&my_local_size, 1, MPI_INT,
+ non_uniform_map, 1, MPI_INT, leader_comm);
+ for(i=0; i < leader_group_size; i++) {
+ if(non_uniform_map[0] != non_uniform_map[i]) {
+ is_uniform = 0;
+ break;
+ }
+ }
+ if(!is_uniform && smpi_comm_is_uniform(comm)){
+ comm->non_uniform_map= non_uniform_map;
+ }else{
+ xbt_free(non_uniform_map);
+ }
+ comm->is_uniform=is_uniform;
+ }
+ smpi_coll_tuned_bcast_mpich(&(comm->is_uniform),1, MPI_INT, 0, comm_intra );
+
+ if(smpi_privatize_global_variables){ //we need to switch here, as the called function may silently touch global variables
+ smpi_switch_data_segment(smpi_process_index());
+ }
+ // Are the ranks blocked ? = allocated contiguously on the SMP nodes
+ int is_blocked=1;
+ int prev=smpi_group_rank(smpi_comm_group(comm), smpi_group_index(smpi_comm_group(comm_intra), 0));
+ for (i=1; i<my_local_size; i++){
+ int this=smpi_group_rank(smpi_comm_group(comm),smpi_group_index(smpi_comm_group(comm_intra), i));
+ if(this!=prev+1){
+ is_blocked=0;
+ break;
+ }
+ prev = this;
+ }
+
+ int global_blocked;
+ smpi_mpi_allreduce(&is_blocked, &(global_blocked), 1,
+ MPI_INT, MPI_LAND, comm);
+
+ if(MPI_COMM_WORLD==SMPI_UNINITIALIZED || comm==MPI_COMM_WORLD){
+ if(smpi_comm_rank(comm)==0){
+ comm->is_blocked=global_blocked;
+ }
+ }else{
+ comm->is_blocked=global_blocked;
+ }
+ xbt_free(leader_list);
+
+ if(replaying==1)
+ smpi_process_set_replaying(1);
+}
+
+int smpi_comm_attr_delete(MPI_Comm comm, int keyval){
+ smpi_comm_key_elem elem = xbt_dict_get_or_null_ext(smpi_comm_keyvals, (const char*)&keyval, sizeof(int));
+ if(!elem)
+ return MPI_ERR_ARG;
+ if(elem->delete_fn!=MPI_NULL_DELETE_FN){
+ void * value;
+ int flag;
+ if(smpi_comm_attr_get(comm, keyval, &value, &flag)==MPI_SUCCESS){
+ int ret = elem->delete_fn(comm, keyval, value, &flag);
+ if(ret!=MPI_SUCCESS) return ret;
+ }
+ }
+ if(comm->attributes==NULL)
+ return MPI_ERR_ARG;
+
+ xbt_dict_remove_ext(comm->attributes, (const char*)&keyval, sizeof(int));
+ return MPI_SUCCESS;
+}
+
+int smpi_comm_attr_get(MPI_Comm comm, int keyval, void* attr_value, int* flag){
+ smpi_comm_key_elem elem = xbt_dict_get_or_null_ext(smpi_comm_keyvals, (const char*)&keyval, sizeof(int));
+ if(!elem)
+ return MPI_ERR_ARG;
+ xbt_ex_t ex;
+ if(comm->attributes==NULL){
+ *flag=0;
+ return MPI_SUCCESS;
+ }
+ TRY {
+ *(void**)attr_value = xbt_dict_get_ext(comm->attributes, (const char*)&keyval, sizeof(int));
+ *flag=1;
+ }
+ CATCH(ex) {
+ *flag=0;
+ xbt_ex_free(ex);
+ }
+ return MPI_SUCCESS;
+}
+
+int smpi_comm_attr_put(MPI_Comm comm, int keyval, void* attr_value){
+ if(!smpi_comm_keyvals)
+ smpi_comm_keyvals = xbt_dict_new();
+ smpi_comm_key_elem elem = xbt_dict_get_or_null_ext(smpi_comm_keyvals, (const char*)&keyval, sizeof(int));
+ if(!elem )
+ return MPI_ERR_ARG;
+ int flag;
+ void* value;
+ smpi_comm_attr_get(comm, keyval, &value, &flag);
+ if(flag && elem->delete_fn!=MPI_NULL_DELETE_FN){
+ int ret = elem->delete_fn(comm, keyval, value, &flag);
+ if(ret!=MPI_SUCCESS) return ret;
+ }
+ if(comm->attributes==NULL)
+ comm->attributes=xbt_dict_new();
+
+ xbt_dict_set_ext(comm->attributes, (const char*)&keyval, sizeof(int), attr_value, NULL);
+ return MPI_SUCCESS;
+}
+
+int smpi_comm_keyval_create(MPI_Comm_copy_attr_function* copy_fn, MPI_Comm_delete_attr_function* delete_fn, int* keyval, void* extra_state){
+
+ if(!smpi_comm_keyvals)
+ smpi_comm_keyvals = xbt_dict_new();
+
+ smpi_comm_key_elem value = (smpi_comm_key_elem) xbt_new0(s_smpi_mpi_comm_key_elem_t,1);
+
+ value->copy_fn=copy_fn;
+ value->delete_fn=delete_fn;
+
+ *keyval = comm_keyval_id;
+ xbt_dict_set_ext(smpi_comm_keyvals, (const char*)keyval, sizeof(int),(void*)value, NULL);
+ comm_keyval_id++;
+ return MPI_SUCCESS;
+}
+
+int smpi_comm_keyval_free(int* keyval){
+ smpi_comm_key_elem elem = xbt_dict_get_or_null_ext(smpi_comm_keyvals, (const char*)keyval, sizeof(int));
+ if(!elem){
+ return MPI_ERR_ARG;
+ }
+ xbt_dict_remove_ext(smpi_comm_keyvals, (const char*)keyval, sizeof(int));
+ xbt_free(elem);
+ return MPI_SUCCESS;
}