Proxy ping [PATCH] PR fortran/100136 - ICE, regression, using flag -fcheck=pointer

[Harald Anlauf <anlauf@gmx.de>]

[-- Attachment #1: Type: text/plain, Size: 554 bytes --]

Dear all,

said PR was addressed by Jose in

  https://gcc.gnu.org/pipermail/fortran/2021-April/055949.html

but unfortunately his patch was never reviewed.

IMHO the patch is mostly fine, with one small exception that
it should use POINTER_TYPE_P (TREE_TYPE (tmp)), see PR and
attached adjusted patch.

Regtested on x86_64-pc-linux-gnu.

Since this fixes a 11/12/13 regression on valid code, I'd like
to commit this one on these branches, but would be glad to see
an OK or LGTM from somebody else.  (Jose says he's too busy
currently.)

Thanks,
Harald


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: PR100136.diff --]
[-- Type: text/x-patch, Size: 2617 bytes --]

From 3ce0b852ddd91151a23baf8301b7aea5cc7d7ea0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jos=C3=A9=20Rui=20Faustino=20de=20Sousa?=
 <jrfsousa@gmail.com>
Date: Sun, 4 Sep 2022 21:53:09 +0200
Subject: [PATCH] Fortran: Fix ICE with -fcheck=pointer [PR100136]

gcc/fortran/ChangeLog:

	PR fortran/100136
	* trans-expr.cc (gfc_conv_procedure_call): Add handling of pointer
	expressions.

gcc/testsuite/ChangeLog:

	PR fortran/100136
	* gfortran.dg/PR100136.f90: New test.
---
 gcc/fortran/trans-expr.cc              |  7 ++---
 gcc/testsuite/gfortran.dg/PR100136.f90 | 39 ++++++++++++++++++++++++++
 2 files changed, 42 insertions(+), 4 deletions(-)
 create mode 100644 gcc/testsuite/gfortran.dg/PR100136.f90

diff --git a/gcc/fortran/trans-expr.cc b/gcc/fortran/trans-expr.cc
index 13c3e7df45f..7895d034610 100644
--- a/gcc/fortran/trans-expr.cc
+++ b/gcc/fortran/trans-expr.cc
@@ -7220,16 +7220,15 @@ gfc_conv_procedure_call (gfc_se * se, gfc_symbol * sym,
 	      else
 		goto end_pointer_check;

+	      tmp = parmse.expr;
 	      if (fsym && fsym->ts.type == BT_CLASS)
 		{
-		  tmp = build_fold_indirect_ref_loc (input_location,
-						      parmse.expr);
+		  if (POINTER_TYPE_P (TREE_TYPE (tmp)))
+		    tmp = build_fold_indirect_ref_loc (input_location, tmp);
 		  tmp = gfc_class_data_get (tmp);
 		  if (GFC_DESCRIPTOR_TYPE_P (TREE_TYPE (tmp)))
 		    tmp = gfc_conv_descriptor_data_get (tmp);
 		}
-	      else
-		tmp = parmse.expr;

 	      /* If the argument is passed by value, we need to strip the
 		 INDIRECT_REF.  */
diff --git a/gcc/testsuite/gfortran.dg/PR100136.f90 b/gcc/testsuite/gfortran.dg/PR100136.f90
new file mode 100644
index 00000000000..922af4aecc3
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/PR100136.f90
@@ -0,0 +1,39 @@
+! { dg-do run }
+! { dg-options "-fcheck=pointer" }
+! { dg-shouldfail "Argument not allocated" }
+! { dg-output "Fortran runtime error: Allocatable actual argument 'c_init2' is not allocated" }
+!
+! Tests fix for PR100136
+!
+! Test cut down from PR58586
+!
+
+module test_pr58586_mod
+  implicit none
+
+  type :: a
+  end type
+
+  type :: c
+     type(a), allocatable :: a
+  end type
+
+contains
+
+  subroutine add_class_c (d)
+    class(c), value :: d
+  end subroutine
+
+  class(c) function c_init2()
+    allocatable :: c_init2
+  end function
+
+end module test_pr58586_mod
+
+program test_pr58586
+  use test_pr58586_mod
+
+  ! This needs to execute, to see whether the segfault at runtime is resolved
+  call add_class_c(c_init2())
+
+end program
--
2.35.3