[PATCH] c: Fix up error-recovery on functions initialized as variables [PR109412]

[PATCH] c: Fix up error-recovery on functions initialized as variables [PR109412]

[Jakub Jelinek]

Hi!

The change to allow empty initializers in C broke error-recovery on the
following testcase.  We are emitting function %qD is initialized like a
variable error early; if the initializer is non-empty, we just emit
another error that the initializer is invalid.  Previously if it was empty,
we'd emit another error that scalar is being initialized by empty
initializer (not really correct), but now we instead just try to
build_zero_cst for the FUNCTION_TYPE and ICE on it.

The following patch just emits the same diagnostics for the empty
initializers as we emit for the non-empty ones.

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

2023-04-14  Jakub Jelinek  <jakub@redhat.com>

	PR c/107682
	PR c/109412
	* c-typeck.cc (pop_init_level): If constructor_type is FUNCTION_TYPE,
	reject empty initializer as invalid.

	* gcc.dg/pr109412.c: New test.

--- gcc/c/c-typeck.cc.jj	2023-03-29 08:32:35.747720324 +0200
+++ gcc/c/c-typeck.cc	2023-04-13 11:37:30.353762300 +0200
@@ -9374,6 +9374,11 @@ pop_init_level (location_t loc, int impl
 	{
 	  if (constructor_erroneous || constructor_type == error_mark_node)
 	    ret.value = error_mark_node;
+	  else if (TREE_CODE (constructor_type) == FUNCTION_TYPE)
+	    {
+	      error_init (loc, "invalid initializer");
+	      ret.value = error_mark_node;
+	    }
 	  else if (TREE_CODE (constructor_type) == POINTER_TYPE)
 	    /* Ensure this is a null pointer constant in the case of a
 	       'constexpr' object initialized with {}.  */
--- gcc/testsuite/gcc.dg/pr109412.c.jj	2023-04-13 12:06:27.820694502 +0200
+++ gcc/testsuite/gcc.dg/pr109412.c	2023-04-13 12:06:07.561986811 +0200
@@ -0,0 +1,20 @@
+/* PR c/107682 */
+/* PR c/109412 */
+/* { dg-do compile } */
+/* { dg-options "" } */
+
+char bar () = {};	/* { dg-error "function 'bar' is initialized like a variable" } */
+			/* { dg-error "invalid initializer" "" { target *-*-* } .-1 } */
+			/* { dg-message "near initialization for 'bar'" "" { target *-*-* } .-2 } */
+char baz () = { 1 };	/* { dg-error "function 'baz' is initialized like a variable" } */
+			/* { dg-error "invalid initializer" "" { target *-*-* } .-1 } */
+			/* { dg-message "near initialization for 'baz'" "" { target *-*-* } .-2 } */
+void
+foo ()
+{
+  int qux () = {};	/* { dg-error "function 'qux' is initialized like a variable" } */
+			/* { dg-error "invalid initializer" "" { target *-*-* } .-1 } */
+			/* { dg-message "near initialization for 'qux'" "" { target *-*-* } .-2 } */
+  int corge () = { 1 };	/* { dg-error "function 'corge' is initialized like a variable" } */
+			/* { dg-error "invalid initializer" "" { target *-*-* } .-1 } */
+}			/* { dg-message "near initialization for 'corge'" "" { target *-*-* } .-2 } */

	Jakub

Re: [PATCH] c: Fix up error-recovery on functions initialized as variables [PR109412]

[Joseph Myers]

On Fri, 14 Apr 2023, Jakub Jelinek via Gcc-patches wrote:

> Hi!
> 
> The change to allow empty initializers in C broke error-recovery on the
> following testcase.  We are emitting function %qD is initialized like a
> variable error early; if the initializer is non-empty, we just emit
> another error that the initializer is invalid.  Previously if it was empty,
> we'd emit another error that scalar is being initialized by empty
> initializer (not really correct), but now we instead just try to
> build_zero_cst for the FUNCTION_TYPE and ICE on it.
> 
> The following patch just emits the same diagnostics for the empty
> initializers as we emit for the non-empty ones.
> 
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

OK.

-- 
Joseph S. Myers
joseph@codesourcery.com