public inbox for fortran@gcc.gnu.org
 help / color / mirror / Atom feed
From: Tobias Burnus <tobias@codesourcery.com>
To: gcc-patches <gcc-patches@gcc.gnu.org>,
	fortran <fortran@gcc.gnu.org>, Jakub Jelinek <jakub@redhat.com>
Subject: [Patch] Fortran/OpenMP: Improve associate-name diagnostic [PR103039]
Date: Fri, 18 Mar 2022 17:27:02 +0100	[thread overview]
Message-ID: <6e5a585d-9876-cc66-413d-3cad381ddbdf@codesourcery.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 1270 bytes --]

SELECT TYPE, SELECT RANK and ASSOCIATE have
   associate-name => selector
and create a pointer to the selector.

GCC was fixed to handle CLASS properly in
   class(t) :: var
   !$omp ... firstprivate(var)
As a side effect, firstprivate(assoc_name) now also gets
handled that way, effectively trying to firstprivate(selector)
which should be shared...

While firstprivate(var) does not appear explicitly, it gets
added via gfc_omp_predetermined_sharing.

I went for the simple solution and handle it only
in gfortran's ctor/dtor.

An alternative would be to set OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
which is currently only set for C++'s __for_end / __for_range
and then later process it in ctor/dtor. I am not sure whether that's
really best and what's the best way to propagate it. One way would
be to create and use OMP_CLAUSE_DEFAULT_FIRSTPRIVATE_NO_REFERENCE.

OK as is (simple version) – or is a fuller version better. If so,
suggestion how to do this best?

Tobias
-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[-- Attachment #2: fix-assoc-name-priv.diff --]
[-- Type: text/x-patch, Size: 3893 bytes --]

Fortran/OpenMP: Fix privatization of associated names

gfc_omp_predetermined_sharing cases the associate-name pointer variable
to be OMP_CLAUSE_DEFAULT_FIRSTPRIVATE, which is fine. However, the associated
selector is shared. Thus, the target of associate-name pointer should not get
copied. (It was before but because of gfc_omp_privatize_by_reference returning
false, the selector was not only wrongly copied but this was also not done
properly.)

gcc/fortran/ChangeLog:

	PR fortran/103039
	* trans-openmp.cc (gfc_omp_clause_copy_ctor, gfc_omp_clause_dtor):
	Only privatize pointer for associate names.

libgomp/ChangeLog:

	PR fortran/103039
	* testsuite/libgomp.fortran/associate4.f90: New test.

 gcc/fortran/trans-openmp.cc                      | 10 +++
 libgomp/testsuite/libgomp.fortran/associate4.f90 | 92 ++++++++++++++++++++++++
 2 files changed, 102 insertions(+)

diff --git a/gcc/fortran/trans-openmp.cc b/gcc/fortran/trans-openmp.cc
index fad76a4791f..101924f5e76 100644
--- a/gcc/fortran/trans-openmp.cc
+++ b/gcc/fortran/trans-openmp.cc
@@ -808,6 +808,11 @@ gfc_omp_clause_copy_ctor (tree clause, tree dest, tree src)
   gcc_assert (OMP_CLAUSE_CODE (clause) == OMP_CLAUSE_FIRSTPRIVATE
 	      || OMP_CLAUSE_CODE (clause) == OMP_CLAUSE_LINEAR);
 
+  /* Privatize pointer, only; cf. gfc_omp_predetermined_sharing. */
+  if (DECL_P (OMP_CLAUSE_DECL (clause))
+      && GFC_DECL_ASSOCIATE_VAR_P (OMP_CLAUSE_DECL (clause)))
+    return build2 (MODIFY_EXPR, TREE_TYPE (dest), dest, src);
+
   if (DECL_ARTIFICIAL (OMP_CLAUSE_DECL (clause))
       && DECL_LANG_SPECIFIC (OMP_CLAUSE_DECL (clause))
       && GFC_DECL_SAVED_DESCRIPTOR (OMP_CLAUSE_DECL (clause)))
@@ -1321,6 +1326,11 @@ gfc_omp_clause_dtor (tree clause, tree decl)
   tree type = TREE_TYPE (decl), tem;
   tree decl_type = TREE_TYPE (OMP_CLAUSE_DECL (clause));
 
+  /* Only pointer was privatized; cf. gfc_omp_clause_copy_ctor. */
+  if (DECL_P (OMP_CLAUSE_DECL (clause))
+      && GFC_DECL_ASSOCIATE_VAR_P (OMP_CLAUSE_DECL (clause)))
+    return NULL_TREE;
+
   if (DECL_ARTIFICIAL (OMP_CLAUSE_DECL (clause))
       && DECL_LANG_SPECIFIC (OMP_CLAUSE_DECL (clause))
       && GFC_DECL_SAVED_DESCRIPTOR (OMP_CLAUSE_DECL (clause)))
diff --git a/libgomp/testsuite/libgomp.fortran/associate4.f90 b/libgomp/testsuite/libgomp.fortran/associate4.f90
new file mode 100644
index 00000000000..f0949b5530d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.fortran/associate4.f90
@@ -0,0 +1,92 @@
+! PR fortran/51722
+
+module m
+implicit none
+
+contains
+
+subroutine seltype
+
+type :: a
+   integer :: p = 2
+end type a
+
+type, extends(a) :: b
+   integer :: cnt = 0
+end type b
+
+integer :: k, s
+class(a), pointer :: x
+
+allocate(a :: x)
+s = 0
+select type (y => x)
+class is (a)
+!$omp parallel do default(shared) private(k) reduction(+:s)
+   do k = 1,10
+      s = s + k*y%p
+   end do
+!$omp end parallel do
+end select
+
+if (s /= 110) error stop
+deallocate(x)
+allocate(b :: x)
+
+s = 0
+select type (y => x)
+class is (b)
+!$omp parallel do default(shared) private(k) reduction(+:s)
+   do k = 1,10
+      s = s + k*y%p
+!$omp atomic update
+      y%cnt = y%cnt + 2
+   end do
+!$omp end parallel do
+if (s /= 110) error stop
+if (y%p /= 2) error stop
+if (y%cnt /= 10*2) error stop
+end select
+
+deallocate(x)
+
+end subroutine seltype
+
+subroutine assoc
+
+type :: b
+   integer :: r = 3
+end type b
+
+type :: a
+   integer :: p = 2
+   class(b), pointer :: u => null()
+end type a
+
+integer :: k, s
+class(a), pointer :: x
+
+s = 0
+allocate(a :: x)
+allocate(b :: x%u)
+
+associate(f => x%u)
+!$omp parallel do default(shared) private(k) reduction(+:s)
+   do k = 1,10
+      s = s + k*f%r
+   end do
+!$omp end parallel do
+end associate
+
+deallocate(x%u)
+deallocate(x)
+
+if (s /= 165) error stop
+end subroutine assoc
+end module m
+
+use m
+implicit none (type, external)
+call seltype
+call assoc
+end

             reply	other threads:[~2022-03-18 16:27 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-18 16:27 Tobias Burnus [this message]
2022-03-18 16:38 ` Jakub Jelinek
  -- strict thread matches above, loose matches on Subject: below --
2022-03-18 13:15 Tobias Burnus
2022-03-18 13:21 ` Jakub Jelinek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6e5a585d-9876-cc66-413d-3cad381ddbdf@codesourcery.com \
    --to=tobias@codesourcery.com \
    --cc=fortran@gcc.gnu.org \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jakub@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).