OffTarget

Observed issues › gcc/gcc-acc-use-device-allocatable-rejected

Appears not to supportRefused at compile time · Fortran · OpenACC · Open

gfortran: an allocatable array in host_data use_device is rejected ("ALLOCATABLE object ... in USE_DEVICE clause")

Legality: independently confirmed. 2 independent readings find the program legal and the result required. How this is checked

GCC 16.2's Fortran front end rejects an allocatable array in the use_device clause of host_data ("ALLOCATABLE object 'a' in USE_DEVICE clause"). OpenACC asks only that the item be a variable or an array. CCE and NVHPC compile and run the programs.

Affected variants

CompilerMI300A gfx942
unified memory (XNACK 1)
MI300A gfx942
copy (XNACK 0)
MI250X gfx90a
copy (XNACK 0)
MI300X gfx942
copy (XNACK 0)
MI350X gfx950
copy (XNACK 0)
A100 sm_80
separate memory
A100 sm_80
managed memory
AFAR 24.3.0n/an/an/an/an/an/an/a
CCE 19.0.0correctcorrectunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 20.0.2correctcorrectunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 21.0.3correctcorrectunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 21.0.4correctcorrectunknown
not run
unknown
not run
unknown
not run
n/an/a
NVHPC 25.11n/an/an/an/an/acorrectcorrect
NVHPC 26.9n/an/an/an/an/acorrectcorrect
GCC 16.2.0not supportednot supportednot supportednot supportednot supportednot supportedn/a

"behaves incorrectly" / "not supported": this issue, confirmed in up to four environments (another build directory, another node, a repeat). "fails another way": a different failure, not counted as this issue. "unknown": not run there, or the runs disagree.

Messages

ResultCompiler and GPUMessage
not supportedGCC 16.2.0, MI300A unified
GCC 16.2.0, MI300A copy
GCC 16.2.0, MI250X
GCC 16.2.0, MI300X
GCC 16.2.0, MI350X
GCC 16.2.0, A100 separate
show
site-gcc--gcc-acc-use-device-allocatable-rejected.f90:76:32:
   76 |     !$acc host_data use_device(a, b)
      |                                1
Error: ALLOCATABLE object ‘a’ in USE_DEVICE clause at (1)
site-gcc--gcc-acc-use-device-allocatable-rejected.f90:76:35:
   76 |     !$acc host_data use_device(a, b)
      |                                   1
Error: ALLOCATABLE object ‘b’ in USE_DEVICE clause at (1)

Reproduce

GCC 16.2.0, MI300A unified

export LD_LIBRARY_PATH=$GCC_16_2/lib64:/opt/rocm-7.2.1/lib
$GCC_16_2/bin/gfortran -fopenacc -foffload=amdgcn-amdhsa -foffload-options=amdgcn-amdhsa=-march=gfx942 -O2 repro.f90 -o repro
HSA_XNACK=1 ./repro

GCC 16.2.0, A100 separate

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=$GCC_16_2/lib64
$GCC_16_2/bin/gfortran -fopenacc -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 repro.f90 -o repro
./repro

CCE 19.0.0, MI300A unified (runs correctly)

module load cce/19.0.0 craype-accel-amd-gfx942 rocm/6.4.3
module unload cray-mpich
ftn -h acc -O2 repro.f90 -o repro
HSA_XNACK=1 ./repro
Program (83 lines)
! # Finding (GCC 16.2): an allocatable array in use_device is rejected (Fortran); the reproducer is the catalog program
!   acc34-clause-use_device
!
! - toolchains: gcc_g90 (GCC 16.2.0, AMD Instinct MI250X gfx90a, Tioga) and gcc_sm80 (GCC 16.2.0,
!   NVIDIA A100 sm_80, wingtip); separate memory only; -O2; OMP_TARGET_OFFLOAD=mandatory
! - build: gfortran -fopenacc -foffload=amdgcn-amdhsa -foffload-options=amdgcn-amdhsa=-march=gfx90a -O2 (gcc_g90);
!   -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 (gcc_sm80)
! - failure: rejected: ALLOCATABLE object .* in USE_DEVICE clause
! - observed: `Error: ALLOCATABLE object 'a' in USE_DEVICE clause at (1)` on both GPUs (the host_data and deviceptr programs pass
!   their allocatable arrays the same way).
! - expected: compiles; `RESULT ran_on_host=0 check=pass`
! - spec: "A var in a use_device clause must be the name of a variable or array" (OpenACC 3.4 §2.8 l.2230); the restriction names no
!   attribute, and an allocatable array is an array.
! - GCC 16.2 documentation (libgomp.texi "Enabling OpenACC"): "A complete description of all OpenACC directives accepted may be
!   found in the OpenACC Application Programming Interface manual, version 2.6"; _OPENACC is 201711 (2.6). No list of OpenACC gaps
!   is documented.
! - other families: CCE 19/20/21 (MI300A, MI250X) and NVHPC 25.11/26.9 compile and run all three programs correctly.
! - gate and confirmer: 6 cells (3 gcc_g90, 3 gcc_sm80), each confirmed in the gate's four environments (a-d;
!   results/gcc/gate-tioga/ and gate-wingtip/ verdicts.jsonl); confirmer verdicts (results/gcc/confirm/batch.gcc_*.jsonl):
!   compiler_bug (R11-compile-rejection) 6
! - catalog programs (3): acc34-clause-deviceptr, acc34-clause-use_device, acc34-directive-host_data
! - legality: the catalog programs are legal by construction (offload.engine.gen, c_gen and acc_gen
!   check every program against the catalog rules and the offload scope as it is built; no
!   compiler). GCC is a compiler under test only: never a legality witness, here or for another
!   family's case (catalog/machines.yaml under_test_only).
! Generated by compliance.gen (OpenACC 3.4, free-form Fortran). Do not edit.
! id: acc34-clause-use_device
! primary atom: acc.clause.use_device
! legal: yes
! observable: yes
! ran_on_host: inside the compute region of interest (acc_on_device)
! evidence: copy_mode_only
! needs: none
! std: f2008
module gen_mod
  use openacc
  implicit none
  contains
    subroutine dev_twice(x, y, m, hf)
      integer, intent(in) :: m
      integer :: x(m), y(m), hf
      integer :: k
      !$acc parallel loop deviceptr(x, y) copy(hf)
      do k = 1, m
        if (k == 1) hf = merge(1, 0, acc_on_device(acc_device_host))
        y(k) = 2*x(k) + 1
      end do
    end subroutine dev_twice
end module gen_mod

program gen_prog
  use openacc, only: acc_device_host, acc_on_device
  use gen_mod
  implicit none
  integer, parameter :: n = 64
  integer, allocatable, target :: a(:), b(:)
  integer :: ref(n)
  integer :: i, host_flag
  logical :: ok
  logical :: pres, shared_mem
  integer :: zc(1)
  allocate(a(n), b(n))
  do i = 1, n
    a(i) = i
    b(i) = 0
    ref(i) = 2*i + 1
  end do
  host_flag = 1
  pres = .true.
  zc = 0
  !$acc serial copyin(zc)
    zc(1) = 1
  !$acc end serial
  shared_mem = zc(1) == 1
  !$acc data copyin(a) copyout(b)
    !$acc host_data use_device(a, b)
      call dev_twice(a, b, n, host_flag)
    !$acc end host_data
  !$acc end data
  ok = all(b == ref) .and. pres
  print '(a,i0,a,a)', 'RESULT ran_on_host=', host_flag, ' check=', merge('pass', 'fail', ok)
  deallocate(a, b)
end program gen_prog

Source in the repository

Something to add? Dispute its legality · Report it fixed in a newer release · Report a mistake on this page

Expected and observed

Expected

check=pass

Observed where it behaves incorrectly

site-gcc--gcc-acc-use-device-allocatable-rejected.f90:76:32:
   76 |     !$acc host_data use_device(a, b)
      |                                1
Error: ALLOCATABLE object ‘a’ in USE_DEVICE clause at (1)
site-gcc--gcc-acc-use-device-allocatable-rejected.f90:76:35:
   76 |     !$acc host_data use_device(a, b)
      |                                   1
Error: ALLOCATABLE object ‘b’ in USE_DEVICE clause at (1)

What the specification requires

The specification text this program relies on and the compiler is expected to support.

OpenACC 3.4 §2.8

A var in a use_device clause must be the name of a variable or array

Evidence

Other compilers run it correctly

CCE 19.0.0, CCE 20.0.2, CCE 21.0.3, CCE 21.0.4, NVHPC 25.11, NVHPC 26.9