diff mbox series

Fortran: Delay vtab generation until after parsing [PR92587]

Message ID 6b20a856-d983-5e37-47d2-6749a1a6cfcf@codesourcery.com
State New
Headers show
Series Fortran: Delay vtab generation until after parsing [PR92587] | expand

Commit Message

Tobias Burnus Dec. 16, 2020, 10:23 p.m. UTC 
Calling gfc_find_vtab during resolution during parsing
comes too early for finalizers. This patch just moves it
to the resolution, which seems to be the simplest solution.

(I tried way more complicated ones which do not work; thus,
I settled for the simple solution ...)

OK for GCC 9/10/mainline?

Tobias

PS: I was wondering about FORALL (which is resolved separately),
but as "f = b" does not work (missing loop index) and as
"f(i)%a = b" is broken (no reallocation, no __copy call,
→ https://gcc.gnu.org/PR98336 ) I think no action is needed
for now.

-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany
Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter

Comments

Thomas Koenig Dec. 17, 2020, 9:02 a.m. UTC | #1 
Hi Tobias,

> Calling gfc_find_vtab during resolution during parsing
> comes too early for finalizers. This patch just moves it
> to the resolution, which seems to be the simplest solution.

Ah, finalizers - one area where we still are deficient WRT Fortran 2003.
Thanks for doing things in this area!

> (I tried way more complicated ones which do not work; thus,
> I settled for the simple solution ...)

Certainly sounds like a good idea to do so.

> OK for GCC 9/10/mainline?

OK. Just one remark: Could you make the test into a run-time test
to check that the finalizers are called correctly?

Best regards

	Thomas
Tobias Burnus Dec. 17, 2020, 9:41 a.m. UTC | #2 
Hi Thomas, hi all,

On 17.12.20 10:02, Thomas Koenig via Fortran wrote:
> OK. Just one remark: Could you make the test into a run-time test
> to check that the finalizers are called correctly?

Done – see attachment.

Thanks for the quick review!

Tobias

-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany
Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter
Thomas Koenig Dec. 17, 2020, 10:58 a.m. UTC | #3 
Hi Tobias,

> Thanks for the quick review!

It helps when I'm on holiday :-)

Best regards

	Thomas
diff mbox series

Patch

Fortran: Delay vtab generation until after parsing [PR92587]

gcc/fortran/ChangeLog:

	PR fortran/92587
	* match.c (gfc_match_assignment): Move gfc_find_vtab call from here ...
	* resolve.c (gfc_resolve_code): ... to here.

gcc/testsuite/ChangeLog:

	PR fortran/92587
	* gfortran.dg/finalize_37.f90: New test.

 gcc/fortran/match.c                          |  3 --
 gcc/fortran/resolve.c                        |  3 ++
 gcc/testsuite/gfortran.dg/finalize_37.f90    | 51 ++++++++++++++++++++++++++++
 4 files changed, 80 insertions(+), 3 deletions(-)

diff --git a/gcc/fortran/match.c b/gcc/fortran/match.c
index bee73e7b008..c13fe96ed33 100644
--- a/gcc/fortran/match.c
+++ b/gcc/fortran/match.c
@@ -1389,9 +1389,6 @@  gfc_match_assignment (void)
 
   gfc_check_do_variable (lvalue->symtree);
 
-  if (lvalue->ts.type == BT_CLASS)
-    gfc_find_vtab (&rvalue->ts);
-
   return MATCH_YES;
 }
 
diff --git a/gcc/fortran/resolve.c b/gcc/fortran/resolve.c
index 05a5e43c90b..1da7ba4d031 100644
--- a/gcc/fortran/resolve.c
+++ b/gcc/fortran/resolve.c
@@ -11896,6 +11896,9 @@  start:
 	  if (!t)
 	    break;
 
+	  if (code->expr1->ts.type == BT_CLASS)
+	   gfc_find_vtab (&code->expr2->ts);
+
 	  /* Remove a GFC_ISYM_CAF_GET inserted for a coindexed variable on
 	     the LHS.  */
 	  if (code->expr1->expr_type == EXPR_FUNCTION
diff --git a/gcc/testsuite/gfortran.dg/finalize_37.f90 b/gcc/testsuite/gfortran.dg/finalize_37.f90
new file mode 100644
index 00000000000..3f872633fdd
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/finalize_37.f90
@@ -0,0 +1,51 @@ 
+! { dg-do compile }
+! { dg-additional-options "-fdump-tree-original" }
+!
+! PR fortran/92587
+!
+
+module m
+   implicit none (type, external)
+   type t2
+   contains
+      final :: fini
+   end type
+   type t3
+      type(t2) :: a
+   end type
+   type, extends(t3) :: t4
+   end type
+   class(t4), allocatable :: y
+   class(t4), allocatable :: z
+contains
+   subroutine sub
+      y = z
+   end
+   subroutine fini(x)
+     type(t2) :: x
+   end
+end
+
+module m2
+  use m
+  implicit none (type, external)
+  type, extends(t3) :: t5
+  end type
+  type, extends(t3) :: t6
+    contains
+      final :: fin2
+  end type
+contains
+  subroutine bar(x, y, z)
+    class(t4), allocatable, intent(out) :: x
+    class(t5), allocatable, intent(out) :: y
+    class(t6), allocatable, intent(out) :: z
+  end
+  subroutine fin2 (x)
+    type(t6) :: x
+  end
+end  
+
+! { dg-final { scan-tree-dump "__final_m_T2 \\\(struct" "original" } }
+! { dg-final { scan-tree-dump "__final_m_T3 \\\(struct" "original" } }
+! { dg-final { scan-tree-dump "__final_m2_T6 \\\(struct" "original" } }