attribs: Don't diagnose attribute exclusions during error recovery [PR94705]

On the following testcase GCC ICEs, because last_decl is error_mark_node,
and diag_attr_exclusions assumes that if it is not NULL, it must be a decl.

The following patch just doesn't diagnose attribute exclusions if the
other decl is erroneous (and thus we've already reported errors for it).

2020-04-23  Jakub Jelinek  <jakub@redhat.com>

	PR c/94705
	* attribs.c (decl_attribute): Don't diagnose attribute exclusions
	if last_decl is error_mark_node or has such a TREE_TYPE.

	* gcc.dg/pr94705.c: New test.

gcc/ChangeLog

@@ -1,3 +1,9 @@
+2020-04-23  Jakub Jelinek  <jakub@redhat.com>
+
+	PR c/94705
+	* attribs.c (decl_attribute): Don't diagnose attribute exclusions
+	if last_decl is error_mark_node or has such a TREE_TYPE.
+
 2020-04-22  Felix Yang  <felix.yang@huawei.com>
 
 	PR target/94678

gcc/attribs.c

@@ -677,7 +677,8 @@ decl_attributes (tree *node, tree attributes, int flags,
 	 reject incompatible attributes.  */
       bool built_in = flags & ATTR_FLAG_BUILT_IN;
       if (spec->exclude
-	  && (flag_checking || !built_in))
+	  && (flag_checking || !built_in)
+	  && !error_operand_p (last_decl))
 	{
 	  /* Always check attributes on user-defined functions.
 	     Check them on built-ins only when -fchecking is set.

gcc/testsuite/ChangeLog

@@ -1,3 +1,8 @@
+2020-04-23  Jakub Jelinek  <jakub@redhat.com>
+
+	PR c/94705
+	* gcc.dg/pr94705.c: New test.
+
 2020-04-22  Patrick Palka  <ppalka@redhat.com>
 
 	PR c++/94719

gcc/testsuite/gcc.dg/pr94705.c

@@ -0,0 +1,13 @@
+/* PR c/94705 */
+/* { dg-do compile } */
+/* { dg-options "" } */
+
+void foo ();
+
+int
+bar (void)
+{
+  foo (baz);	/* { dg-error "'baz' undeclared" } */
+		/* { dg-message "only once" "" { target *-*-* } .-1 } */
+  void __attribute__ ((noinline)) baz (void);
+}