(In reply to Tom de Vries from comment #0)
> @r257131, with GOMP_NVPTX_JIT=-O0, driver version 384.111, geforce 710, and
> cuda 8.0, kernel version 4.13.0-32-generic:

Same result without GOMP_NVPTX_JIT=-O0.

And same result with GOMP_NVPTX_JIT=-O0 on driver version 384.111, quadro m1200, cuda 7.5, kernel version 4.13.0-31-generic.

Minimal version:
...
! { dg-do run }                                                                                                    

module vars
  implicit none
  integer z
  !$acc declare create (z)                                                                                         
end module vars

program main
  use vars
  use openacc
  implicit none

  if (acc_is_present (z) .neqv. .true.) call abort

end program
...

Bisect points to commit r257065:
...
commit d9c7c3e3f6eb4621f929474e0ba44e7d61584431 (HEAD)
Author: pault <pault@138bc75d-0d04-0410-961f-82ee72b054a4>
Date:   Thu Jan 25 19:09:40 2018 +0000

    2018-25-01  Paul Thomas  <pault@gcc.gnu.org>
    
            PR fortran/37577
            * array.c (gfc_match_array_ref): If standard earlier than F2008
            it is an error if the reference dimension is greater than 7.
            libgfortran.h : Increase GFC_MAX_DIMENSIONS to 15. Change the
            dtype masks and shifts accordingly.
            * trans-array.c (gfc_conv_descriptor_dtype): Use the dtype
            type node to check the field.
            (gfc_conv_descriptor_dtype): Access the rank field of dtype.
            (duplicate_allocatable_coarray): Access the rank field of the
            dtype descriptor rather than the dtype itself.
            * trans-expr.c (get_scalar_to_descriptor_type): Store the type
            of 'scalar' on entry and use its TREE_TYPE if it is ARRAY_TYPE
            (ie. a character).
            (gfc_conv_procedure_call): Pass TREE_OPERAND (tmp,0) to
            get_scalar_to_descriptor_type if the actual expression is a
            constant.
            (gfc_trans_structure_assign): Assign the rank directly to the
            dtype rank field.
            * trans-intrinsic.c (gfc_conv_intrinsic_rank): Cast the result
            to default integer kind.
            (gfc_conv_intrinsic_sizeof): Obtain the element size from the
            'elem_len' field of the dtype.
            * trans-io.c (gfc_build_io_library_fndecls): Replace
            gfc_int4_type_node with dtype_type_node where necessary.
            (transfer_namelist_element): Use gfc_get_dtype_rank_type for
            scalars.
            * trans-types.c : Provide 'get_dtype_type_node' to acces the
            dtype_type_node and, if necessary, build it.
            The maximum size of an array element is now determined by the
            maximum value of size_t.
            Update the description of the array descriptor, including the
            type def for the dtype_type.
            (gfc_get_dtype_rank_type): Build a constructor for the dtype.
            Distinguish RECORD_TYPEs that are BT_DERIVED or BT_CLASS.
            (gfc_get_array_descriptor_base): Change the type of the dtype
            field to dtype_type_node.
            (gfc_get_array_descr_info): Get the offset to the rank field of
            the dtype.
            * trans-types.h : Add a prototype for 'get_dtype_type_node ()'.
            * trans.h : Define the indices of the dtype fields.
    
    2018-25-01  Paul Thomas  <pault@gcc.gnu.org>
    
            PR fortran/37577
            * gfortran.dg/coarray_18.f90: Allow dimension 15 for F2008.
            * gfortran.dg/coarray_lib_this_image_2.f90: Change 'array1' to
            'array01' in the tree dump comparison.
            * gfortran.dg/coarray_lib_token_4.f90: Likewise.
            * gfortran.dg/inline_sum_1.f90: Similar - allow two digits.
            * gfortran.dg/rank_1.f90: Allow dimension 15 for F2008.
    
    2018-25-01  Paul Thomas  <pault@gcc.gnu.org>
    
            PR fortran/37577
            * caf/single.c (_gfortran_caf_failed_images): Access the 'type'
            and 'elem_len' fields of the dtype instead of the shifts.
            (_gfortran_caf_stopped_images): Likewise.
            * intrinsics/associated.c (associated): Compare the 'type' and
            'elem_len' fields instead of the dtype.
            * caf/date_and_time.c : Access the dtype fields rather using
            shifts and masks.
            * io/transfer.c (transfer_array ): Comment on item count.
            (set_nml_var,st_set_nml_var): Change dtype type and use fields.
            (st_set_nml_dtio_var): Likewise.
            * libgfortran.h : Change definition of GFC_ARRAY_DESCRIPTOR and
            add a typedef for the dtype_type. Change the GFC_DTYPE_* macros
            to access the dtype fields.
    
    
    
    git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@257065 138bc75d-0d04-0410-961f-82ee72b054a4
...

Will reconfirm with full build.

(In reply to Tom de Vries from comment #3)
> Will reconfirm with full build.

Did clean builds of r257064 and r257065. Minimal test passes at r257064, fails at r257065. Confirmed.

Hmm. Probably this failure would have been picked up by libgomp-plugin-host_nonshm.

(In reply to Tom de Vries from comment #5)
> Hmm. Probably this failure would have been picked up by
> libgomp-plugin-host_nonshm.

Hi Tom,

Although my patch caused it, I am not in a position to pick this one up - I am unable to reproduce it and do not have access to the target you are using. In addition, libgomp and all its workings are, as far as I am concerned, one of the arcane mysteries of gcc.

I am perfectly happy to put in some time to help but I do not even know where to start.

Sorry

Paul

(In reply to Paul Thomas from comment #6)
> (In reply to Tom de Vries from comment #5)
> > Hmm. Probably this failure would have been picked up by
> > libgomp-plugin-host_nonshm.
> 
> Hi Tom,
> 
> Although my patch caused it, I am not in a position to pick this one up - I
> am unable to reproduce it and do not have access to the target you are
> using. In addition, libgomp and all its workings are, as far as I am
> concerned, one of the arcane mysteries of gcc.
> 
> I am perfectly happy to put in some time to help but I do not even know
> where to start.

I may have found a way to reproduce the problem without libgomp (disclaimer: I have no idea whether this testcase if valid fortran or not).

Consider this testcase:
...
  implicit none
  integer(kind=4) z

  call foo (z)

contains
  subroutine foo (a)
    type (*), dimension (..), contiguous :: a
    print *, sizeof(a)
  end subroutine foo

end program
...

compiled like this:
...
$ gfortran -O2 test.f90
...

gives:
...
$ ./a.out 
                    8
...

Previously, the testcase printed 4 instead of 8.

> > am unable to reproduce it and do not have access to the target you are
> > using.

> I may have found a way to reproduce the problem without libgomp

Forgot to mention: on x86_64, so you should be able to reproduce this.

Ha! Thanks...

In the main programme:
    struct array00_integer(kind=4) desc.3;

    desc.3.dtype = {.elem_len=8, .rank=0, .type=11};
    desc.3.data = (void * restrict) &z;
    foo (&desc.3);

At least foo is reporting the correct result. I'm onto it.

Paul


On 30 January 2018 at 10:59, vries at gcc dot gnu.org
<gcc-bugzilla@gcc.gnu.org> wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84088
>
> --- Comment #8 from Tom de Vries <vries at gcc dot gnu.org> ---
>> > am unable to reproduce it and do not have access to the target you are
>> > using.
>
>> I may have found a way to reproduce the problem without libgomp
>
> Forgot to mention: on x86_64, so you should be able to reproduce this.
>
> --
> You are receiving this mail because:
> You are on the CC list for the bug.

I'll take this and deal with it this evening.

The problem lies in trans-expr.c(gfc_conv_scalar_to_descriptor), which was touched in the patch as was the code around the call to it from gfc_conv_procedure_call.

You will note that not only is elem_len = 8 in comment #9 but the type is 11, which is also wrong.

Thanks for the testcase with which I can reproduce the problem.

Paul

Author: pault
Date: Wed Jan 31 20:28:35 2018
New Revision: 257262

URL: https://gcc.gnu.org/viewcvs?rev=257262&root=gcc&view=rev
Log:
2018-01-31  Paul Thomas  <pault@gcc.gnu.org>

	PR fortran/84088
	* trans-expr.c (gfc_conv_procedure_call): If the parm expr is
	an address expression passed to an assumed rank dummy, convert
	to an indirect reference.

2018-01-31  Paul Thomas  <pault@gcc.gnu.org>

	PR fortran/84088
	* gfortran.dg/pr84088.f90 : New test.


Added:
    trunk/gcc/testsuite/gfortran.dg/pr84088.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-expr.c
    trunk/gcc/testsuite/ChangeLog

Patch committed with non-openacc test-case from comment 7.

Marking resolved-fixed.