Re: [PATCH] c++: fix ICE with is_really_empty_class [PR110106]

[Marek Polacek <polacek@redhat.com>]

On Fri, Jul 21, 2023 at 01:44:17PM -0400, Jason Merrill wrote:
> On 7/20/23 17:58, Marek Polacek wrote:
> > On Thu, Jul 20, 2023 at 03:51:32PM -0400, Marek Polacek wrote:
> > > On Thu, Jul 20, 2023 at 02:37:07PM -0400, Jason Merrill wrote:
> > > > On 7/20/23 14:13, Marek Polacek wrote:
> > > > > On Wed, Jul 19, 2023 at 10:11:27AM -0400, Patrick Palka wrote:
> > > > > > On Tue, 18 Jul 2023, Marek Polacek via Gcc-patches wrote:
> > > > > > 
> > > > > > > Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk and branches?
> > > > > > 
> > > > > > Looks reasonable to me.
> > > > > 
> > > > > Thanks.
> > > > > > Though I wonder if we could also fix this by not checking potentiality
> > > > > > at all in this case?  The problematic call to is_rvalue_constant_expression
> > > > > > happens from cp_parser_constant_expression with 'allow_non_constant' != 0
> > > > > > and with 'non_constant_p' being a dummy out argument that comes from
> > > > > > cp_parser_functional_cast, so the result of is_rvalue_constant_expression
> > > > > > is effectively unused in this case, and we should be able to safely elide
> > > > > > it when 'allow_non_constant && non_constant_p == nullptr'.
> > > > > 
> > > > > Sounds plausible.  I think my patch could be applied first since it
> > > > > removes a tiny bit of code, then I can hopefully remove the flag below,
> > > > > then maybe go back and optimize the call to is_rvalue_constant_expression.
> > > > > Does that sound sensible?
> > > > > 
> > > > > > Relatedly, ISTM the member cp_parser::non_integral_constant_expression_p
> > > > > > is also effectively unused and could be removed?
> > > > > 
> > > > > It looks that way.  Seems it's only used in cp_parser_constant_expression:
> > > > > 10806   if (allow_non_constant_p)
> > > > > 10807     *non_constant_p = parser->non_integral_constant_expression_p;
> > > > > but that could be easily replaced by a local var.  I'd be happy to see if
> > > > > we can actually do away with it.  (I wonder why it was introduced and when
> > > > > it actually stopped being useful.)
> > > > 
> > > > It was for the C++98 notion of constant-expression, which was more of a
> > > > parser-level notion, and has been supplanted by the C++11 version.  I'm
> > > > happy to remove it, and therefore remove the is_rvalue_constant_expression
> > > > call.
> > > 
> > > Wonderful.  I'll do that next.
> > 
> > I found a use of parser->non_integral_constant_expression_p:
> > finish_id_expression_1 can set it to true which then makes
> > a difference in cp_parser_constant_expression in C++98.  In
> > cp_parser_constant_expression we set n_i_c_e_p to false, call
> > cp_parser_assignment_expression in which finish_id_expression_1
> > sets n_i_c_e_p to true, then back in cp_parser_constant_expression
> > we skip the cxx11 block, and set *non_constant_p to true.  If I
> > remove n_i_c_e_p, we lose that.  This can be seen in init/array60.C.
> 
> Sure, we would need to use the C++11 code for C++98 mode, which is likely
> fine but is more uncertain.
> 
> It's probably simpler to just ignore n_i_c_e_p for C++11 and up, along with
> Patrick's suggestion of allowing null non_constant_p with true
> allow_non_constant_p.

Something like this, then?  I see that cp_parser_initializer_clause et al
offer further opportunities (because they sometimes use a dummy too) but
this should be a good start.

Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?

-- >8 --
It's pointless to call *_rvalue_constant_expression when we're not using
the result.  Also apply some drive-by cleanups.

gcc/cp/ChangeLog:

	* parser.cc (cp_parser_constant_expression): Allow non_constant_p to be
	nullptr even when allow_non_constant_p is true.  Don't call
	_rvalue_constant_expression when not necessary.  Move local variable
	declarations closer to their first use.
	(cp_parser_static_assert): Don't pass a dummy down to
	cp_parser_constant_expression.
---
 gcc/cp/parser.cc | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

diff --git a/gcc/cp/parser.cc b/gcc/cp/parser.cc
index 5e2b5cba57e..efaa806f107 100644
--- a/gcc/cp/parser.cc
+++ b/gcc/cp/parser.cc
@@ -10734,11 +10734,6 @@ cp_parser_constant_expression (cp_parser* parser,
 			       bool *non_constant_p /* = NULL */,
 			       bool strict_p /* = false */)
 {
-  bool saved_integral_constant_expression_p;
-  bool saved_allow_non_integral_constant_expression_p;
-  bool saved_non_integral_constant_expression_p;
-  cp_expr expression;
-
   /* It might seem that we could simply parse the
      conditional-expression, and then check to see if it were
      TREE_CONSTANT.  However, an expression that is TREE_CONSTANT is
@@ -10757,10 +10752,12 @@ cp_parser_constant_expression (cp_parser* parser,
      will fold this operation to an INTEGER_CST for `3'.  */
 
   /* Save the old settings.  */
-  saved_integral_constant_expression_p = parser->integral_constant_expression_p;
-  saved_allow_non_integral_constant_expression_p
+  bool saved_integral_constant_expression_p
+    = parser->integral_constant_expression_p;
+  bool saved_allow_non_integral_constant_expression_p
     = parser->allow_non_integral_constant_expression_p;
-  saved_non_integral_constant_expression_p = parser->non_integral_constant_expression_p;
+  bool saved_non_integral_constant_expression_p
+    = parser->non_integral_constant_expression_p;
   /* We are now parsing a constant-expression.  */
   parser->integral_constant_expression_p = true;
   parser->allow_non_integral_constant_expression_p
@@ -10780,6 +10777,7 @@ cp_parser_constant_expression (cp_parser* parser,
      For example, cp_parser_initializer_clauses uses this function to
      determine whether a particular assignment-expression is in fact
      constant.  */
+  cp_expr expression;
   if (strict_p)
     expression = cp_parser_conditional_expression (parser);
   else
@@ -10789,7 +10787,8 @@ cp_parser_constant_expression (cp_parser* parser,
     = saved_integral_constant_expression_p;
   parser->allow_non_integral_constant_expression_p
     = saved_allow_non_integral_constant_expression_p;
-  if (cxx_dialect >= cxx11)
+  if (cxx_dialect >= cxx11
+      && (!allow_non_constant_p || non_constant_p))
     {
       /* Require an rvalue constant expression here; that's what our
 	 callers expect.  Reference constant expressions are handled
@@ -10803,7 +10802,7 @@ cp_parser_constant_expression (cp_parser* parser,
       if (!is_const && !allow_non_constant_p)
 	require_rvalue_constant_expression (decay);
     }
-  if (allow_non_constant_p)
+  if (allow_non_constant_p && non_constant_p)
     *non_constant_p = parser->non_integral_constant_expression_p;
   parser->non_integral_constant_expression_p
     = saved_non_integral_constant_expression_p;
@@ -16400,12 +16399,11 @@ cp_parser_linkage_specification (cp_parser* parser, tree prefix_attr)
    If MEMBER_P, this static_assert is a class member.  */
 
 static void
-cp_parser_static_assert(cp_parser *parser, bool member_p)
+cp_parser_static_assert (cp_parser *parser, bool member_p)
 {
   cp_expr condition;
   location_t token_loc;
   tree message;
-  bool dummy;
 
   /* Peek at the `static_assert' token so we can keep track of exactly
      where the static assertion started.  */
@@ -16430,7 +16428,7 @@ cp_parser_static_assert(cp_parser *parser, bool member_p)
   condition =
     cp_parser_constant_expression (parser,
                                    /*allow_non_constant_p=*/true,
-                                   /*non_constant_p=*/&dummy);
+				   /*non_constant_p=*/nullptr);
 
   if (cp_lexer_peek_token (parser->lexer)->type == CPP_CLOSE_PAREN)
     {

base-commit: 6e424febfbcb27c21a7fe3a137e614765f9cf9d2
-- 
2.41.0