diff mbox series

Fortran: Use OpenACC's acc_on_device builtin, fix OpenMP' __builtin_is_initial_device: Fix effective-target keyword in 'libgomp.oacc-fortran/acc_on_device-2.f90'

Message ID 87h69f84wq.fsf@euler.schwinge.ddns.net
State New
Headers show
Series Fortran: Use OpenACC's acc_on_device builtin, fix OpenMP' __builtin_is_initial_device: Fix effective-target keyword in 'libgomp.oacc-fortran/acc_on_device-2.f90' | expand

Commit Message

Thomas Schwinge Oct. 14, 2024, 8:28 a.m. UTC 
Hi!

On 2024-10-14T10:23:56+0200, I wrote:
> On 2024-10-13T10:21:01+0200, Tobias Burnus <tburnus@baylibre.com> wrote:
>> Now pushed as r15-4298-g3269a722b7a036.

>>>>> Tested on x86-64 without and with offloading configured, running 
>>>>> with nvptx offloading.

I see an UNRESOLVED:

    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_host=1 -DACC_MEM_SHARED=1 -foffload=disable  -O   scan-tree-dump-not optimized "acc_on_device"
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_host=1 -DACC_MEM_SHARED=1 -foffload=disable  -O   scan-tree-dump-times gimple "acc_on_device" 1
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_host=1 -DACC_MEM_SHARED=1 -foffload=disable  -O  (test for excess errors)
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none  -O   at line 37 (test for warnings, line 36)
    +UNRESOLVED: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none  -O   scan-nvptx-none-offload-tree-dump-not optimized "acc_on_device"
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none  -O   scan-tree-dump-not optimized "acc_on_device"
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none  -O   scan-tree-dump-times gimple "acc_on_device" 1
    +PASS: libgomp.oacc-fortran/acc_on_device-2.f90 -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none  -O  (test for excess errors)

>> --- /dev/null
>> +++ b/libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90
>> @@ -0,0 +1,40 @@
>> +! { dg-do link }
>> +
>> +! Check whether 'acc_on_device()' is properly compile-time optimized. */
>> +
>> +! { dg-additional-options "-fdump-tree-gimple -fdump-tree-optimized" }
>> +! { dg-additional-options -foffload-options=-fdump-tree-optimized { target { offload_device_nvptx || offload_target_amdgcn } } }
>> +
>> +! { dg-final { scan-tree-dump-times "acc_on_device" 1 "gimple" } }
>> +
>> +! { dg-final { scan-tree-dump-not "acc_on_device" "optimized" } }
>> +
>> +! { dg-final { only_for_offload_target amdgcn-amdhsa scan-offload-tree-dump-not "acc_on_device" "optimized" { target offload_target_amdgcn } } }
>> +! { dg-final { only_for_offload_target nvptx-none scan-offload-tree-dump-not "acc_on_device" "optimized" { target offload_target_nvptx } } }

Pushed to trunk branch commit c3774b2e2d7d00ad9f9f6fce10aa6bc872bd951f
"Fortran: Use OpenACC's acc_on_device builtin, fix OpenMP' __builtin_is_initial_device: Fix effective-target keyword in 'libgomp.oacc-fortran/acc_on_device-2.f90'",
see attached.


Grüße
 Thomas


>> +
>> +
>> +module m
>> +   integer :: xxxx
>> +   !$acc declare device_resident(xxxx)
>> +contains
>> +  subroutine set_var
>> +    !$acc routine
>> +    use openacc
>> +    implicit none (type, external)
>> +    if (acc_on_device(acc_device_host)) then
>> +      xxxx = 1234
>> +    else
>> +      xxxx = 4242
>> +    end if
>> +  end
>> +end module m
>> +
>> +
>> +program main
>> +  use m
>> +  call set_var
>> +  !$acc serial
>> +    ! { dg-warning "using 'vector_length \\(32\\)', ignoring 1" "" { target openacc_nvidia_accel_selected } .-1 }
>> +    call set_var
>> +  !$acc end serial
>> +end
diff mbox series

Patch

From c3774b2e2d7d00ad9f9f6fce10aa6bc872bd951f Mon Sep 17 00:00:00 2001
From: Thomas Schwinge <tschwinge@baylibre.com>
Date: Mon, 14 Oct 2024 10:26:13 +0200
Subject: [PATCH] Fortran: Use OpenACC's acc_on_device builtin, fix OpenMP'
 __builtin_is_initial_device: Fix effective-target keyword in
 'libgomp.oacc-fortran/acc_on_device-2.f90'

The test case 'libgomp.oacc-fortran/acc_on_device-2.f90' added in
commit 3269a722b7a03613e9c4e2862bc5088c4a17cc11
"Fortran: Use OpenACC's acc_on_device builtin, fix OpenMP' __builtin_is_initial_device"
had a mismatch between dump file production and its scanning; the former needs
to use 'offload_target_nvptx' (like 'offload_target_amdgcn'), not
'offload_device_nvptx'.

	PR testsuite/82250
	libgomp/
	* testsuite/libgomp.oacc-fortran/acc_on_device-2.f90: Fix
	effective-target keyword.
---
 libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90 b/libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90
index 39d4357dd55..306555cfccf 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90
+++ b/libgomp/testsuite/libgomp.oacc-fortran/acc_on_device-2.f90
@@ -3,7 +3,7 @@ 
 ! Check whether 'acc_on_device()' is properly compile-time optimized. */
 
 ! { dg-additional-options "-fdump-tree-gimple -fdump-tree-optimized" }
-! { dg-additional-options -foffload-options=-fdump-tree-optimized { target { offload_device_nvptx || offload_target_amdgcn } } }
+! { dg-additional-options -foffload-options=-fdump-tree-optimized { target { offload_target_nvptx || offload_target_amdgcn } } }
 
 ! { dg-final { scan-tree-dump-times "acc_on_device" 1 "gimple" } }
 
-- 
2.34.1