Skip to content

Fix/nb leaks #8926

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 2 commits into from
Closed

Fix/nb leaks #8926

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8d9b63d
Release the temporary arrays for nonblocking collectives.
bosilca Oct 6, 2021
76128f8
Fix hcoll compile error.
awlauria Apr 14, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion ompi/include/ompi/memchecker.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -308,7 +308,7 @@ static inline int memchecker_request(MPI_Request *request)
opal_memchecker_base_isdefined (&(*request)->req_status._ucount, sizeof(size_t));
#endif

opal_memchecker_base_isdefined ((void*)&(*request)->req_complete, sizeof(volatile _Bool));
opal_memchecker_base_isdefined ((void*)&(*request)->req_complete, sizeof(volatile void*));
opal_memchecker_base_isdefined ((void*)&(*request)->req_state, sizeof(volatile ompi_request_state_t));
opal_memchecker_base_isdefined (&(*request)->req_persistent, sizeof(_Bool));
opal_memchecker_base_isdefined (&(*request)->req_f_to_c_index, sizeof(int));
Expand Down
67 changes: 41 additions & 26 deletions ompi/mca/coll/base/coll_base_util.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2020 The University of Tennessee and The University
* Copyright (c) 2004-2021 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
Expand Down Expand Up @@ -127,14 +127,28 @@ int ompi_rounddown(int num, int factor)
return num * factor; /* floor(num / factor) * factor */
}

static void release_objs_callback(struct ompi_coll_base_nbc_request_t *request) {
if (NULL != request->data.objs.objs[0]) {
OBJ_RELEASE(request->data.objs.objs[0]);
request->data.objs.objs[0] = NULL;
/**
* Release all objects and arrays stored into the nbc_request.
* The release_arrays are temporary memory to stored the values
* converted from Fortran, and should dissapear in same time as the
* request itself.
*/
static void
release_objs_callback(struct ompi_coll_base_nbc_request_t *request)
{
if (NULL != request->data.refcounted.objs.objs[0]) {
OBJ_RELEASE(request->data.refcounted.objs.objs[0]);
request->data.refcounted.objs.objs[0] = NULL;
}
if (NULL != request->data.refcounted.objs.objs[1]) {
OBJ_RELEASE(request->data.refcounted.objs.objs[1]);
request->data.refcounted.objs.objs[1] = NULL;
}
if (NULL != request->data.objs.objs[1]) {
OBJ_RELEASE(request->data.objs.objs[1]);
request->data.objs.objs[1] = NULL;
for(int i = 0; i < OMPI_REQ_NB_RELEASE_ARRAYS; i++ ) {
if (NULL == request->data.release_arrays[i])
break;
free(request->data.release_arrays[i]);
request->data.release_arrays[i] = NULL;
}
}

Expand Down Expand Up @@ -168,12 +182,12 @@ int ompi_coll_base_retain_op( ompi_request_t *req, ompi_op_t *op,
}
if (!ompi_op_is_intrinsic(op)) {
OBJ_RETAIN(op);
request->data.op.op = op;
request->data.refcounted.op.op = op;
retain = true;
}
if (!ompi_datatype_is_predefined(type)) {
OBJ_RETAIN(type);
request->data.op.datatype = type;
request->data.refcounted.op.datatype = type;
retain = true;
}
if (OPAL_UNLIKELY(retain)) {
Expand Down Expand Up @@ -207,12 +221,12 @@ int ompi_coll_base_retain_datatypes( ompi_request_t *req, ompi_datatype_t *stype
}
if (NULL != stype && !ompi_datatype_is_predefined(stype)) {
OBJ_RETAIN(stype);
request->data.types.stype = stype;
request->data.refcounted.types.stype = stype;
retain = true;
}
if (NULL != rtype && !ompi_datatype_is_predefined(rtype)) {
OBJ_RETAIN(rtype);
request->data.types.rtype = rtype;
request->data.refcounted.types.rtype = rtype;
retain = true;
}
if (OPAL_UNLIKELY(retain)) {
Expand All @@ -237,21 +251,21 @@ static void release_vecs_callback(ompi_coll_base_nbc_request_t *request) {
} else {
scount = rcount = OMPI_COMM_IS_INTER(comm)?ompi_comm_remote_size(comm):ompi_comm_size(comm);
}
if (NULL != request->data.vecs.stypes) {
if (NULL != request->data.refcounted.vecs.stypes) {
for (int i=0; i<scount; i++) {
if (NULL != request->data.vecs.stypes[i]) {
OMPI_DATATYPE_RELEASE_NO_NULLIFY(request->data.vecs.stypes[i]);
if (NULL != request->data.refcounted.vecs.stypes[i]) {
OMPI_DATATYPE_RELEASE_NO_NULLIFY(request->data.refcounted.vecs.stypes[i]);
}
}
request->data.vecs.stypes = NULL;
request->data.refcounted.vecs.stypes = NULL;
}
if (NULL != request->data.vecs.rtypes) {
if (NULL != request->data.refcounted.vecs.rtypes) {
for (int i=0; i<rcount; i++) {
if (NULL != request->data.vecs.rtypes[i]) {
OMPI_DATATYPE_RELEASE_NO_NULLIFY(request->data.vecs.rtypes[i]);
if (NULL != request->data.refcounted.vecs.rtypes[i]) {
OMPI_DATATYPE_RELEASE_NO_NULLIFY(request->data.refcounted.vecs.rtypes[i]);
}
}
request->data.vecs.rtypes = NULL;
request->data.refcounted.vecs.rtypes = NULL;
}
}

Expand Down Expand Up @@ -304,8 +318,8 @@ int ompi_coll_base_retain_datatypes_w( ompi_request_t *req,
}
}
if (OPAL_UNLIKELY(retain)) {
request->data.vecs.stypes = (ompi_datatype_t **) stypes;
request->data.vecs.rtypes = (ompi_datatype_t **) rtypes;
request->data.refcounted.vecs.stypes = (ompi_datatype_t **) stypes;
request->data.refcounted.vecs.rtypes = (ompi_datatype_t **) rtypes;
if (req->req_persistent) {
request->cb.req_free = req->req_free;
req->req_free = free_vecs_callback;
Expand All @@ -319,15 +333,16 @@ int ompi_coll_base_retain_datatypes_w( ompi_request_t *req,
return OMPI_SUCCESS;
}

static void nbc_req_cons(ompi_coll_base_nbc_request_t *req)
static void nbc_req_constructor(ompi_coll_base_nbc_request_t *req)
{
req->cb.req_complete_cb = NULL;
req->req_complete_cb_data = NULL;
req->data.objs.objs[0] = NULL;
req->data.objs.objs[1] = NULL;
req->data.refcounted.objs.objs[0] = NULL;
req->data.refcounted.objs.objs[1] = NULL;
req->data.release_arrays[0] = NULL;
}

OBJ_CLASS_INSTANCE(ompi_coll_base_nbc_request_t, ompi_request_t, nbc_req_cons, NULL);
OBJ_CLASS_INSTANCE(ompi_coll_base_nbc_request_t, ompi_request_t, nbc_req_constructor, NULL);

/* File reading functions */
static void skiptonewline (FILE *fptr, int *fileline)
Expand Down
44 changes: 27 additions & 17 deletions ompi/mca/coll/base/coll_base_util.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2020 The University of Tennessee and The University
* Copyright (c) 2004-2021 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2007 High Performance Computing Center Stuttgart,
Expand Down Expand Up @@ -34,6 +34,13 @@

BEGIN_C_DECLS

/**
* The largest array we need to track collective temporary memory. Right now
* the record is for ialltoallw, for the array of send and receive types,
* count and displacements.
*/
#define OMPI_REQ_NB_RELEASE_ARRAYS 7

/**
* Request structure to be returned by non-blocking
* collective operations.
Expand All @@ -45,22 +52,25 @@ struct ompi_coll_base_nbc_request_t {
ompi_request_free_fn_t req_free;
} cb;
void *req_complete_cb_data;
union {
struct {
ompi_op_t *op;
ompi_datatype_t *datatype;
} op;
struct {
ompi_datatype_t *stype;
ompi_datatype_t *rtype;
} types;
struct {
opal_object_t *objs[2];
} objs;
struct {
ompi_datatype_t * const *stypes;
ompi_datatype_t * const *rtypes;
} vecs;
struct {
union {
struct {
ompi_op_t *op;
ompi_datatype_t *datatype;
} op;
struct {
ompi_datatype_t *stype;
ompi_datatype_t *rtype;
} types;
struct {
opal_object_t *objs[2];
} objs;
struct {
ompi_datatype_t * const *stypes;
ompi_datatype_t * const *rtypes;
} vecs;
} refcounted;
void* release_arrays[OMPI_REQ_NB_RELEASE_ARRAYS];
} data;
};

Expand Down
4 changes: 2 additions & 2 deletions ompi/mca/coll/hcoll/coll_hcoll_rte.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -352,8 +352,8 @@ static void* get_coll_handle(void)
ompi_req->super.req_state = OMPI_REQUEST_ACTIVE;
ompi_req->super.req_free = request_free;
ompi_req->super.req_type = OMPI_REQUEST_COLL;
ompi_req->data.objs.objs[0] = NULL;
ompi_req->data.objs.objs[1] = NULL;
ompi_req->data.refcounted.objs.objs[0] = NULL;
ompi_req->data.refcounted.objs.objs[1] = NULL;
return (void *)ompi_req;
}

Expand Down
21 changes: 15 additions & 6 deletions ompi/mpi/fortran/mpif-h/allgatherv_init_f.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2005 The University of Tennessee and The University
* Copyright (c) 2004-2021 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
Expand All @@ -23,6 +23,7 @@

#include "ompi/mpi/fortran/mpif-h/bindings.h"
#include "ompi/mpi/fortran/base/constants.h"
#include "ompi/mca/coll/base/coll_base_util.h"

#if OMPI_BUILD_MPI_PROFILING
#if OPAL_HAVE_WEAK_SYMBOLS
Expand Down Expand Up @@ -76,7 +77,7 @@ void ompi_allgatherv_init_f(char *sendbuf, MPI_Fint *sendcount, MPI_Fint *sendty
MPI_Datatype c_sendtype, c_recvtype;
MPI_Request c_request;
MPI_Info c_info;
int size, ierr_c;
int size, idx = 0, ierr_c;
OMPI_ARRAY_NAME_DECL(recvcounts);
OMPI_ARRAY_NAME_DECL(displs);

Expand All @@ -102,8 +103,16 @@ void ompi_allgatherv_init_f(char *sendbuf, MPI_Fint *sendcount, MPI_Fint *sendty
c_recvtype, c_comm, c_info, &c_request);

if (NULL != ierr) *ierr = OMPI_INT_2_FINT(ierr_c);
if (MPI_SUCCESS == ierr_c) *request = PMPI_Request_c2f(c_request);

OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(displs);
if (MPI_SUCCESS == ierr_c) {
*request = PMPI_Request_c2f(c_request);
ompi_coll_base_nbc_request_t* nb_request = (ompi_coll_base_nbc_request_t*)c_request;
if (recvcounts != OMPI_ARRAY_NAME_CONVERT(recvcounts)) {
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(recvcounts);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(displs);
}
nb_request->data.release_arrays[idx] = NULL;
} else {
OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(displs);
}
}
27 changes: 19 additions & 8 deletions ompi/mpi/fortran/mpif-h/alltoallv_init_f.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2005 The University of Tennessee and The University
* Copyright (c) 2004-2021 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
Expand All @@ -23,6 +23,7 @@

#include "ompi/mpi/fortran/mpif-h/bindings.h"
#include "ompi/mpi/fortran/base/constants.h"
#include "ompi/mca/coll/base/coll_base_util.h"

#if OMPI_BUILD_MPI_PROFILING
#if OPAL_HAVE_WEAK_SYMBOLS
Expand Down Expand Up @@ -76,7 +77,7 @@ void ompi_alltoallv_init_f(char *sendbuf, MPI_Fint *sendcounts, MPI_Fint *sdispl
MPI_Datatype c_sendtype, c_recvtype;
MPI_Info c_info;
MPI_Request c_request;
int size, c_ierr;
int size, idx = 0, c_ierr;
OMPI_ARRAY_NAME_DECL(sendcounts);
OMPI_ARRAY_NAME_DECL(sdispls);
OMPI_ARRAY_NAME_DECL(recvcounts);
Expand Down Expand Up @@ -106,10 +107,20 @@ void ompi_alltoallv_init_f(char *sendbuf, MPI_Fint *sendcounts, MPI_Fint *sdispl
OMPI_ARRAY_NAME_CONVERT(rdispls),
c_recvtype, c_comm, c_info, &c_request);
if (NULL != ierr) *ierr = OMPI_INT_2_FINT(c_ierr);
if (MPI_SUCCESS == c_ierr) *request = PMPI_Request_c2f(c_request);

OMPI_ARRAY_FINT_2_INT_CLEANUP(sendcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(sdispls);
OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(rdispls);
if (MPI_SUCCESS == c_ierr) {
*request = PMPI_Request_c2f(c_request);
ompi_coll_base_nbc_request_t* nb_request = (ompi_coll_base_nbc_request_t*)c_request;
if (sendcounts != OMPI_ARRAY_NAME_CONVERT(sendcounts)) {
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(sendcounts);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(sdispls);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(recvcounts);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(rdispls);
}
nb_request->data.release_arrays[idx] = NULL;
} else {
OMPI_ARRAY_FINT_2_INT_CLEANUP(sendcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(sdispls);
OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(rdispls);
}
}
40 changes: 28 additions & 12 deletions ompi/mpi/fortran/mpif-h/alltoallw_init_f.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2005 The University of Tennessee and The University
* Copyright (c) 2004-2021 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
Expand All @@ -23,7 +23,7 @@

#include "ompi/mpi/fortran/mpif-h/bindings.h"
#include "ompi/mpi/fortran/base/constants.h"
#include "ompi/communicator/communicator.h"
#include "ompi/mca/coll/base/coll_base_util.h"

#if OMPI_BUILD_MPI_PROFILING
#if OPAL_HAVE_WEAK_SYMBOLS
Expand Down Expand Up @@ -78,7 +78,7 @@ void ompi_alltoallw_init_f(char *sendbuf, MPI_Fint *sendcounts,
MPI_Datatype *c_sendtypes = NULL, *c_recvtypes;
MPI_Info c_info;
MPI_Request c_request;
int size, c_ierr;
int size, idx = 0, c_ierr;
OMPI_ARRAY_NAME_DECL(sendcounts);
OMPI_ARRAY_NAME_DECL(sdispls);
OMPI_ARRAY_NAME_DECL(recvcounts);
Expand Down Expand Up @@ -117,14 +117,30 @@ void ompi_alltoallw_init_f(char *sendbuf, MPI_Fint *sendcounts,
OMPI_ARRAY_NAME_CONVERT(rdispls),
c_recvtypes, c_comm, c_info, &c_request);
if (NULL != ierr) *ierr = OMPI_INT_2_FINT(c_ierr);
if (MPI_SUCCESS == c_ierr) *request = PMPI_Request_c2f(c_request);

OMPI_ARRAY_FINT_2_INT_CLEANUP(sendcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(sdispls);
OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(rdispls);
if (NULL != c_sendtypes) {
free(c_sendtypes);
if (MPI_SUCCESS == c_ierr) {
*request = PMPI_Request_c2f(c_request);
ompi_coll_base_nbc_request_t* nb_request = (ompi_coll_base_nbc_request_t*)c_request;
nb_request->data.release_arrays[idx++] = c_recvtypes;
if (recvcounts != OMPI_ARRAY_NAME_CONVERT(recvcounts)) {
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(recvcounts);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(rdispls);
}
if (NULL != c_sendtypes) {
nb_request->data.release_arrays[idx++] = c_sendtypes;
if (sendcounts != OMPI_ARRAY_NAME_CONVERT(sendcounts)) {
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(sendcounts);
nb_request->data.release_arrays[idx++] = OMPI_ARRAY_NAME_CONVERT(sdispls);
}
}
nb_request->data.release_arrays[idx] = NULL;
} else {
OMPI_ARRAY_FINT_2_INT_CLEANUP(sendcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(sdispls);
OMPI_ARRAY_FINT_2_INT_CLEANUP(recvcounts);
OMPI_ARRAY_FINT_2_INT_CLEANUP(rdispls);
if (NULL != c_sendtypes) {
free(c_sendtypes);
}
free(c_recvtypes);
}
free(c_recvtypes);
}
Loading