[MPICH] test/mpi/comm/cmfree.c

[Rajeev Thakur]

Correction!  The test is ok as it is. The comm_split is on a different
communicator. And a comm_free is called at the very end of the program if
comm != null. 

Rajeev 

> -----Original Message-----
> From: Rajeev Thakur [mailto:thakur at mcs.anl.gov] 
> Sent: Thursday, July 12, 2007 5:22 PM
> To: 'Toon Knapen'; 'mpich-discuss at mcs.anl.gov'
> Subject: RE: [MPICH] test/mpi/comm/cmfree.c
> 
> Good catch. MPI_Comm_free is collective, so it must be called 
> by everyone. Just adding it to the else statements doesn't 
> work because the communicator is used there after 
> MPI_Comm_split, which is also collective. And you can't 
> interleave collectives. 
> 
> We will fix it somehow or remove the test.
> 
> Rajeev
>  
> 
> > -----Original Message-----
> > From: owner-mpich-discuss at mcs.anl.gov 
> > [mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Toon Knapen
> > Sent: Thursday, July 12, 2007 6:51 AM
> > To: mpich-discuss at mcs.anl.gov
> > Subject: [MPICH] test/mpi/comm/cmfree.c
> > 
> > Hi all,
> > 
> > IMO test/mpi/comm/cmfree.c in mpich2-1.0.5p4 there is one 
> > MPI_Comm_free 
> > missing.
> > 
> > On line 29, there is a `MPI_Comm_dup(MPI_COMM_WORLD,&comm)`. 
> > However the 
> > comm that gets created here is only freed if 'rank == dest' 
> > on line 54. 
> > In all ranks that are not equal to 'dest', the communicator 
> > is never freed.
> > 
> > Of course, this is not an error (just a leak) but wanted to 
> > let you know 
> > nevertheless,
> > 
> > toon
> > 
> > 
>