c++: Fix up handling of references to anon union members in initializers [PR53932]

  Hi!

For anonymous union members we create artificial VAR_DECLs which
have DECL_VALUE_EXPR for the actual COMPONENT_REF.  That works
just fine inside of functions (including global dynamic constructors),
because during gimplification such VAR_DECLs are gimplified as
their DECL_VALUE_EXPR.  This is also done during regimplification.

But references to these artificial vars in DECL_INITIAL expressions
aren't ever replaced by the DECL_VALUE_EXPRs, so we end up either
with link failures like on the testcase below, or worse ICEs with
LTO.

The following patch fixes those during cp_fully_fold_init where we
already walk all the trees (!data->genericize means that
function rather than cp_fold_function).

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2023-01-19  Jakub Jelinek  <jakub@redhat.com>

	PR c++/53932
	* cp-gimplify.cc (cp_fold_r): During cp_fully_fold_init replace
	DECL_ANON_UNION_VAR_P VAR_DECLs with their corresponding
	DECL_VALUE_EXPR.

	* g++.dg/init/pr53932.C: New test.


	Jakub
  

On 1/19/23 15:13, Jakub Jelinek wrote:
> Hi!
> 
> For anonymous union members we create artificial VAR_DECLs which
> have DECL_VALUE_EXPR for the actual COMPONENT_REF.  That works
> just fine inside of functions (including global dynamic constructors),
> because during gimplification such VAR_DECLs are gimplified as
> their DECL_VALUE_EXPR.  This is also done during regimplification.
> 
> But references to these artificial vars in DECL_INITIAL expressions
> aren't ever replaced by the DECL_VALUE_EXPRs, so we end up either
> with link failures like on the testcase below, or worse ICEs with
> LTO.
> 
> The following patch fixes those during cp_fully_fold_init where we
> already walk all the trees (!data->genericize means that
> function rather than cp_fold_function).
> 
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

OK.

> 2023-01-19  Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR c++/53932
> 	* cp-gimplify.cc (cp_fold_r): During cp_fully_fold_init replace
> 	DECL_ANON_UNION_VAR_P VAR_DECLs with their corresponding
> 	DECL_VALUE_EXPR.
> 
> 	* g++.dg/init/pr53932.C: New test.
> 
> --- gcc/cp/cp-gimplify.cc.jj	2023-01-16 11:52:16.065734330 +0100
> +++ gcc/cp/cp-gimplify.cc	2023-01-19 18:13:54.592661735 +0100
> @@ -1010,6 +1010,16 @@ cp_fold_r (tree *stmt_p, int *walk_subtr
>   	}
>         break;
>   
> +    case VAR_DECL:
> +      /* In initializers replace anon union artificial VAR_DECLs
> +	 with their DECL_VALUE_EXPRs, as nothing will do it later.  */
> +      if (DECL_ANON_UNION_VAR_P (stmt) && !data->genericize)
> +	{
> +	  *stmt_p = stmt = unshare_expr (DECL_VALUE_EXPR (stmt));
> +	  break;
> +	}
> +      break;
> +
>       default:
>         break;
>       }
> --- gcc/testsuite/g++.dg/init/pr53932.C.jj	2023-01-19 18:22:24.837231192 +0100
> +++ gcc/testsuite/g++.dg/init/pr53932.C	2023-01-19 18:20:51.776586408 +0100
> @@ -0,0 +1,25 @@
> +// PR c++/53932
> +// { dg-do link }
> +
> +static union { int i; };
> +int &r = i;
> +int s = i;
> +int *t = &i;
> +
> +void
> +foo (int **p, int *q)
> +{
> +  static int &u = i;
> +  static int v = i;
> +  static int *w = &i;
> +  int &x = i;
> +  int y = i;
> +  int *z = &i;
> +  *p = &i;
> +  *q = i;
> +}
> +
> +int
> +main ()
> +{
> +}
> 
> 	Jakub
>
  

On Thu, Jan 19, 2023 at 12:13 PM Jakub Jelinek via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> Hi!
>
> For anonymous union members we create artificial VAR_DECLs which
> have DECL_VALUE_EXPR for the actual COMPONENT_REF.  That works
> just fine inside of functions (including global dynamic constructors),
> because during gimplification such VAR_DECLs are gimplified as
> their DECL_VALUE_EXPR.  This is also done during regimplification.
>
> But references to these artificial vars in DECL_INITIAL expressions
> aren't ever replaced by the DECL_VALUE_EXPRs, so we end up either
> with link failures like on the testcase below, or worse ICEs with
> LTO.
>
> The following patch fixes those during cp_fully_fold_init where we
> already walk all the trees (!data->genericize means that
> function rather than cp_fold_function).
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

I noticed (static) structured bindings has a similar issue but is not
fixed by this because this checks to see if it is an anonymous union
decl.
I filed PR 108474 for that.

Thanks,
Andrew Pinski

>
> 2023-01-19  Jakub Jelinek  <jakub@redhat.com>
>
>         PR c++/53932
>         * cp-gimplify.cc (cp_fold_r): During cp_fully_fold_init replace
>         DECL_ANON_UNION_VAR_P VAR_DECLs with their corresponding
>         DECL_VALUE_EXPR.
>
>         * g++.dg/init/pr53932.C: New test.
>
> --- gcc/cp/cp-gimplify.cc.jj    2023-01-16 11:52:16.065734330 +0100
> +++ gcc/cp/cp-gimplify.cc       2023-01-19 18:13:54.592661735 +0100
> @@ -1010,6 +1010,16 @@ cp_fold_r (tree *stmt_p, int *walk_subtr
>         }
>        break;
>
> +    case VAR_DECL:
> +      /* In initializers replace anon union artificial VAR_DECLs
> +        with their DECL_VALUE_EXPRs, as nothing will do it later.  */
> +      if (DECL_ANON_UNION_VAR_P (stmt) && !data->genericize)
> +       {
> +         *stmt_p = stmt = unshare_expr (DECL_VALUE_EXPR (stmt));
> +         break;
> +       }
> +      break;
> +
>      default:
>        break;
>      }
> --- gcc/testsuite/g++.dg/init/pr53932.C.jj      2023-01-19 18:22:24.837231192 +0100
> +++ gcc/testsuite/g++.dg/init/pr53932.C 2023-01-19 18:20:51.776586408 +0100
> @@ -0,0 +1,25 @@
> +// PR c++/53932
> +// { dg-do link }
> +
> +static union { int i; };
> +int &r = i;
> +int s = i;
> +int *t = &i;
> +
> +void
> +foo (int **p, int *q)
> +{
> +  static int &u = i;
> +  static int v = i;
> +  static int *w = &i;
> +  int &x = i;
> +  int y = i;
> +  int *z = &i;
> +  *p = &i;
> +  *q = i;
> +}

> +
> +int
> +main ()
> +{
> +}
>
>         Jakub
>