[Bug libgomp/92835] New: [OpenACC] Run time error with enter./exit data

[caplanr at predsci dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92835

Bug ID: 92835
   Summary: [OpenACC]  Run time error with enter./exit data
   Product: gcc
   Version: 9.1.0
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: libgomp
  Assignee: unassigned at gcc dot gnu.org
  Reporter: caplanr at predsci dot com
CC: jakub at gcc dot gnu.org
  Target Milestone: ---

I am trying to compile and run my OpenACC code with GNU 9.1.0 branch.

I am getting the following error:

libgomp: >>>> GOACC_enter_exit_data UNHANDLED kind 0x01

What does this mean?  My code works fine with PGI.

[Bug libgomp/92835] [OpenACC] Run time error with enter./exit data

[caplanr at predsci dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92835

--- Comment #2 from Ron  ---
(In reply to Thomas Schwinge from comment #1)
> (In reply to Ron from comment #0)
> > I am trying to compile and run my OpenACC code with GNU 9.1.0 branch.
> > 
> > I am getting the following error:
> 
> Thanks for the report.
> 
> > libgomp:  GOACC_enter_exit_data UNHANDLED kind 0x01
> > 
> > What does this mean?  My code works fine with PGI.
> 
> Huh, that's strange.  That's the data mapping kind used for the OpenACC
> 'copyin' clause, which certainly I see handled in 'GOACC_enter_exit_data'. 
> Can you tell what your source code line is?  You can run your code in GDB,
> with 'break gomp_fatal', and when this hits, then 'backtrace'.

Hi,

It turns out my library path was not pointing to the proper library (it was
pointing to GNU 7).  The code now runs (very slowly).

Can you point me to a simple instruction as to what flags to use during
configuration of the 9 branch to enable NVIDIA support?   Do I have to
separately install nvptx-none?   On compilation, do I need to specify the
offload to nvptx-none or should it automatically compile for NVIDIA?

The wiki is some help but a bit out-dated and not easy to follow.

If these questions are not proper for bugzilla (since they are not "bugs", what
forum do you recommend to post questions such as these?

Thanks

 - Ron

[Bug target/102544] GCN offloading not working for 'amdgcn-amd-amdhsa--gfx906:sramecc+:xnack-'

[caplanr at predsci dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102544

--- Comment #7 from Ron  ---
That last case was using a fortran version of the test code as follows:

program SMALL_TEST_F
use openacc
implicit none
integer :: v

v = 0

!$acc parallel copy(v)
if (acc_on_device(ACC_DEVICE_HOST)) then
v = -1
else
v = 2
end if
!$acc end parallel

print *,'v is',v

end program


compiled with gfortran 10 with the flags:
-g -fopenacc -foffload=amdgcn-amdhsa="-march=gfx906" -foffload=-lgfortran
-foffload=-lm