Fortran : ICE in generic_correspondence PR95584

Message ID 819d9d78-5b63-2420-f78d-4b33c16c7f91@codethink.co.uk
State New
Headers show
Series
  • Fortran : ICE in generic_correspondence PR95584
Related show

Commit Message

Mark Eggleston June 25, 2020, 6:33 a.m.
Please find attached fix for PR95584.  Original patch by Steve Kargl.

OK to commit and backport?

Commit message:

Fortran  : ICE in generic_correspondence PR95584

Output an error for ambiguous interfaces in generic interface
instead of ICE.

2020-06-25  Steven G. Kargl  <kargl@gcc.gnu.org>

gcc/fortran/

     PR fortran/95584
     * interface.c (generic_correspondence): Only use the pointer
     to a symbol if exists.

2020-06-25  Mark Eggleston <markeggleston@gcc.gnu.org>

gcc/testsuite/

     PR fortran/95584
     * gfortran.dg/pr95584.f90: New test.

-- 
https://www.codethink.co.uk/privacy.html

Comments

Mark Eggleston June 30, 2020, 1:02 p.m. | #1
I forgot to use [PATCH] in the subject line. ping!

On 25/06/2020 07:33, Mark Eggleston wrote:
> Please find attached fix for PR95584.  Original patch by Steve Kargl.

>

> OK to commit and backport?

>

> Commit message:

>

> Fortran  : ICE in generic_correspondence PR95584

>

> Output an error for ambiguous interfaces in generic interface

> instead of ICE.

>

> 2020-06-25  Steven G. Kargl  <kargl@gcc.gnu.org>

>

> gcc/fortran/

>

>     PR fortran/95584

>     * interface.c (generic_correspondence): Only use the pointer

>     to a symbol if exists.

>

> 2020-06-25  Mark Eggleston <markeggleston@gcc.gnu.org>

>

> gcc/testsuite/

>

>     PR fortran/95584

>     * gfortran.dg/pr95584.f90: New test.

>

-- 
https://www.codethink.co.uk/privacy.html
Thomas Koenig June 30, 2020, 1:27 p.m. | #2
Hi Mark,


> OK to commit and backport?


OK, fixing such a NULL dererefence could also count as obvious
and simple (if it does the right thing :-)

Regarding backport:  I probably would not do it myself; it's
not a regression, and ICE does not have such a large impact.
However, if you want to do it, feel free.

Thanks for taking this on!

Regards

	Thomas

Patch

From 1205be65bd7c3aff52c2a9a00f298afa7aa0bf08 Mon Sep 17 00:00:00 2001
From: Mark Eggleston <markeggleston@gcc.gnu.org>
Date: Thu, 25 Jun 2020 05:16:50 +0100
Subject: [PATCH] Fortran  : ICE in generic_correspondence PR95584

Output an error for ambiguous interfaces in generic interface
instead of ICE.

2020-06-25  Steven G. Kargl  <kargl@gcc.gnu.org>

gcc/fortran/

	PR fortran/95584
	* interface.c (generic_correspondence): Only use the pointer
	to a symbol if exists.

2020-06-25  Mark Eggleston  <markeggleston@gcc.gnu.org>

gcc/testsuite/

	PR fortran/95584
	* gfortran.dg/pr95584.f90: New test.
---
 gcc/fortran/interface.c               |  2 +-
 gcc/testsuite/gfortran.dg/pr95584.f90 | 16 ++++++++++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/gfortran.dg/pr95584.f90

diff --git a/gcc/fortran/interface.c b/gcc/fortran/interface.c
index b1a75a37b0e..0cc504f4e04 100644
--- a/gcc/fortran/interface.c
+++ b/gcc/fortran/interface.c
@@ -1257,7 +1257,7 @@  generic_correspondence (gfc_formal_arglist *f1, gfc_formal_arglist *f2,
 
   while (f1)
     {
-      if (f1->sym->attr.optional)
+      if (!f1->sym || f1->sym->attr.optional)
 	goto next;
 
       if (p1 && strcmp (f1->sym->name, p1) == 0)
diff --git a/gcc/testsuite/gfortran.dg/pr95584.f90 b/gcc/testsuite/gfortran.dg/pr95584.f90
new file mode 100644
index 00000000000..4504f7b500c
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/pr95584.f90
@@ -0,0 +1,16 @@ 
+! { dg-do compile }
+
+program p
+   interface s
+      subroutine g(x, *)
+      end
+      subroutine h(y, *)
+      end
+   end interface
+end
+
+! { dg-warning "Obsolescent feature: Alternate-return argument" " " { target *-*-* } 5 }
+! { dg-warning "Obsolescent feature: Alternate-return argument" " " { target *-*-* } 7 }
+! { dg-error ".1." " " { target *-*-* } 5 }
+! { dg-error "Ambiguous interfaces in generic interface" " " { target *-*-* } 7 }
+
-- 
2.11.0