Skip to content
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

Determine MPI Data Types in col_on_comm() & dst_on_comm() to prevent displacements overflow. (Fix for #2156) #2157

Merged
merged 4 commits into from
Feb 5, 2025
Merged

Determine MPI Data Types in col_on_comm() & dst_on_comm() to prevent displacements overflow. (Fix for #2156) #2157

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
13bbd1c
remove trailing whitespace
benkirk Jan 17, 2025
05be2ee
Determine MPI Data Types in col_on_comm() & dst_on_comm() to prevent …
benkirk Jan 17, 2025
13b94c2
Only detect overflow in col_on_comm(), report, and exit.
benkirk Jan 18, 2025
0d8def7
Properly define MPI_Datatype dtype.
benkirk Jan 18, 2025
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
73 changes: 54 additions & 19 deletions frame/collect_on_comm.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,11 +36,11 @@
# endif
#endif


int col_on_comm ( int *, int *, void *, int *, void *, int *, int);
int dst_on_comm ( int *, int *, void *, int *, void *, int *, int);

void
void
COLLECT_ON_COMM ( int * comm, int * typesize ,
void * inbuf, int *ninbuf , void * outbuf, int * noutbuf )
{
Expand All @@ -67,8 +67,9 @@ col_on_comm ( int * Fcomm, int * typesize ,
int *displace ;
int noutbuf_loc ;
int root_task ;
MPI_Datatype dtype;
int ierr = -1;
MPI_Comm *comm, dummy_comm ;
int ierr ;

comm = &dummy_comm ;
*comm = MPI_Comm_f2c( *Fcomm ) ;
Expand All @@ -90,28 +91,45 @@ col_on_comm ( int * Fcomm, int * typesize ,
for ( p = 1 , displace[0] = 0 , noutbuf_loc = recvcounts[0] ; p < ntasks ; p++ ) {
displace[p] = displace[p-1]+recvcounts[p-1] ;
noutbuf_loc = noutbuf_loc + recvcounts[p] ;

/* check for overflow: displace is the partial sum of recvcounts, which can overflow for large problems. */
if (displace[p] < 0) {
#ifndef MS_SUA
fprintf(stderr,"%s %d buffer offset overflow!!\n",__FILE__,__LINE__) ;
fprintf(stderr," ---> p = %d,\n ---> displace[%d] = %d,\n ---> typesize = %d\n",
p, p, displace[p], *typesize);
#endif
MPI_Abort(MPI_COMM_WORLD,1) ;
}
}

if ( noutbuf_loc > * noutbuf )
{
#ifndef MS_SUA
fprintf(stderr,"FATAL ERROR: collect_on_comm: noutbuf_loc (%d) > noutbuf (%d)\n",
noutbuf_loc , * noutbuf ) ;
noutbuf_loc , * noutbuf ) ;
fprintf(stderr,"WILL NOT perform the collection operation\n") ;
#endif
MPI_Abort(MPI_COMM_WORLD,1) ;
}

/* multiply everything by the size of the type */
for ( p = 0 ; p < ntasks ; p++ ) {
displace[p] *= *typesize ;
recvcounts[p] *= *typesize ;
}

/* handle different sized data types appropriately. */
ierr = MPI_Type_match_size (MPI_TYPECLASS_REAL, *typesize, &dtype);
if (MPI_SUCCESS != ierr) {
ierr = MPI_Type_match_size (MPI_TYPECLASS_INTEGER, *typesize, &dtype);
if (MPI_SUCCESS != ierr) {
#ifndef MS_SUA
fprintf(stderr,"%s %d FATAL ERROR: unhandled typesize = %d!!\n", __FILE__,__LINE__,*typesize) ;
#endif
MPI_Abort(MPI_COMM_WORLD,1) ;
}
}

ierr = MPI_Gatherv( inbuf , *ninbuf * *typesize , MPI_CHAR ,
outbuf , recvcounts , displace, MPI_CHAR ,
root_task , *comm ) ;
ierr = MPI_Gatherv( inbuf , *ninbuf, dtype,
outbuf , recvcounts , displace, dtype,
root_task , *comm ) ;
#ifndef MS_SUA
if ( ierr != 0 ) fprintf(stderr,"%s %d MPI_Gatherv returns %d\n",__FILE__,__LINE__,ierr ) ;
#endif
Expand Down Expand Up @@ -152,6 +170,8 @@ dst_on_comm ( int * Fcomm, int * typesize ,
int *displace ;
int noutbuf_loc ;
int root_task ;
MPI_Datatype dtype;
int ierr = -1;
MPI_Comm *comm, dummy_comm ;

comm = &dummy_comm ;
Expand All @@ -171,18 +191,34 @@ dst_on_comm ( int * Fcomm, int * typesize ,
for ( p = 1 , displace[0] = 0 , noutbuf_loc = sendcounts[0] ; p < ntasks ; p++ ) {
displace[p] = displace[p-1]+sendcounts[p-1] ;
noutbuf_loc = noutbuf_loc + sendcounts[p] ;

/* check for overflow: displace is the partial sum of sendcounts, which can overflow for large problems. */
if ( (displace[p] < 0) || (noutbuf_loc < 0) ) {
#ifndef MS_SUA
fprintf(stderr,"%s %d buffer offset overflow!!\n",__FILE__,__LINE__) ;
fprintf(stderr," ---> p = %d,\n ---> displace[%d] = %d,\n ---> noutbuf_loc = %d,\n ---> typesize = %d\n",
p, p, displace[p], noutbuf_loc, *typesize);
#endif
MPI_Abort(MPI_COMM_WORLD,1) ;
}
}
}

/* multiply everything by the size of the type */
for ( p = 0 ; p < ntasks ; p++ ) {
displace[p] *= *typesize ;
sendcounts[p] *= *typesize ;
/* handle different sized data types appropriately. */
ierr = MPI_Type_match_size (MPI_TYPECLASS_REAL, *typesize, &dtype);
if (MPI_SUCCESS != ierr) {
ierr = MPI_Type_match_size (MPI_TYPECLASS_INTEGER, *typesize, &dtype);
if (MPI_SUCCESS != ierr) {
#ifndef MS_SUA
fprintf(stderr,"%s %d FATAL ERROR: unhandled typesize = %d!!\n", __FILE__,__LINE__,*typesize) ;
#endif
MPI_Abort(MPI_COMM_WORLD,1) ;
}
}

MPI_Scatterv( inbuf , sendcounts , displace, MPI_CHAR ,
outbuf , *noutbuf * *typesize , MPI_CHAR ,
root_task , *comm ) ;
MPI_Scatterv( inbuf, sendcounts, displace, dtype,
outbuf, *noutbuf, dtype,
root_task, *comm ) ;

free(sendcounts) ;
free(displace) ;
Expand Down Expand Up @@ -241,4 +277,3 @@ rlim_ ()
}
#endif
#endif