Skip to content

defect: MPI crash on exit #795

Open
Open
@sfilippone

Description

@sfilippone

System information including:

  • OpenCoarrays Version: 2.9.2

  • Fortran Compiler: 14.2

  • C compiler used for building lib:

  • Installation method: package manager

  • All flags & options passed to the installer

  • Output of uname -a: Linux lubuntu 6.14.0-15-generic Check Machine Type for Navier-Stokes Tests #15-Ubuntu SMP PREEMPT_DYNAMIC Sun Apr 6 15:05:05 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux

  • MPI library being used: 4.2.1

  • Machine architecture and number of physical cores:

  • Version of CMake:

To help us debug your issue please explain:

What you were trying to do (and why)

Running hello World.
I have been trying to install a recent OpenCoarrays version on both Fedora 42 and Ubuntu, and always get the same result; have tried multiple compiler/MPI combinations including gnu 14.1.0 14.2.0 mpich 4.2.2 4.3.0 always with the same result. If I build with cmake and then do make test, the tests declare PASS, but when I go to the directory containing the executables I get the same as below.
The output below is what I get on ubuntu 24.04 installing the default version from the apt package manager, but the outcome is identical.

Is there anything particularly silly I am doing?
Thanks

What happened (include command output, screenshots, logs, etc.)

Hello from image 1 out of 4
Hello from image 2 out of 4
Hello from image 3 out of 4
Hello from image 4 out of 4
Abort(873553679) on node 0: Fatal error in internal_Comm_free: Other MPI error, error stack:
internal_Comm_free(29495)......: MPI_Comm_free(comm=0x5f953b8910ac) failed
MPIR_Comm_free_impl(823).......:
MPIR_Comm_delete_internal(1267): Communicator (handle=84000002) being freed has 3 unmatched message(s)
Abort(605118223) on node 2: Fatal error in internal_Comm_free: Other MPI error, error stack:
internal_Comm_free(29495)......: MPI_Comm_free(comm=0x5b4303df10ac) failed
MPIR_Comm_free_impl(823).......:
MPIR_Comm_delete_internal(1267): Communicator (handle=84000001) being freed has 3 unmatched message(s)
Abort(403791631) on node 3: Fatal error in internal_Comm_free: Other MPI error, error stack:
internal_Comm_free(29495)......: MPI_Comm_free(comm=0x5aa4d58d90ac) failed
MPIR_Comm_free_impl(823).......:
MPIR_Comm_delete_internal(1267): Communicator (handle=84000001) being freed has 3 unmatched message(s)
Abort(873553679) on node 1: Fatal error in internal_Comm_free: Other MPI error, error stack:
internal_Comm_free(29495)......: MPI_Comm_free(comm=0x64ded0dd10ac) failed
MPIR_Comm_free_impl(823).......:
MPIR_Comm_delete_internal(1267): Communicator (handle=84000001) being freed has 3 unmatched message(s)
Error: Command:
/usr/bin/mpiexec -n 4 ./hello
failed to run.

What you expected to happen

No MPI error on exit

Step-by-step reproduction instructions to reproduce the error/bug

caf -o hello hello.f90
cafrun -np4 ./hello


program main
use iso_fortran_env
implicit none
integer :: me, np, img
integer, parameter :: maxl=80, root=1
character(len=maxl) :: greeting[*]

me = this_image()
np = num_images()
write(greeting,) "Hello from image ",me," out of ",np
sync all
if (me == root) then
do img = 1, np
write(
,*) greeting[img]
end do
end if
sync all
end program main

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions