[Bug c++/98355] New: [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[Bug c++/98355] New: [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[gscfq@t-online.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

            Bug ID: 98355
           Summary: [9/10/11 Regression] ICE in has_attribute, at
                    c-family/c-attribs.c:5628
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gscfq@t-online.de
  Target Milestone: ---

Started to ICE with version r9, roundabout 20181125 :


$ cat z1.cc
struct S { int a; };
template <int> struct T
{
  static_assert (__builtin_has_attribute (((S*)0) -> a, packed));
};


$ g++-11-20201213 -c z1.cc
z1.cc:4:57: internal compiler error: Segmentation fault
    4 |   static_assert (__builtin_has_attribute (((S*)0) -> a, packed));
      |                                                         ^~~~~~
0xc8f5ff crash_signal
        ../../gcc/toplev.c:327
0x841f52 has_attribute(unsigned int, tree_node*, tree_node*, tree_node*
(*)(tree_node*))
        ../../gcc/c-family/c-attribs.c:5628
0x7525ed cp_parser_has_attribute_expression
        ../../gcc/cp/parser.c:8936
0x7525ed cp_parser_unary_expression
        ../../gcc/cp/parser.c:8527
0x72b84f cp_parser_cast_expression
        ../../gcc/cp/parser.c:9712
0x72c082 cp_parser_binary_expression
        ../../gcc/cp/parser.c:9814
0x72c7e0 cp_parser_assignment_expression
        ../../gcc/cp/parser.c:10118
0x72dcbd cp_parser_constant_expression
        ../../gcc/cp/parser.c:10414
0x72f7d2 cp_parser_static_assert
        ../../gcc/cp/parser.c:15319
0x75d23c cp_parser_member_declaration
        ../../gcc/cp/parser.c:25844
0x73774e cp_parser_member_specification_opt
        ../../gcc/cp/parser.c:25712
0x73774e cp_parser_class_specifier_1
        ../../gcc/cp/parser.c:24801
0x739639 cp_parser_class_specifier
        ../../gcc/cp/parser.c:25112
0x739639 cp_parser_type_specifier
        ../../gcc/cp/parser.c:18368
0x73a0b6 cp_parser_decl_specifier_seq
        ../../gcc/cp/parser.c:14990
0x75bd05 cp_parser_single_declaration
        ../../gcc/cp/parser.c:30321
0x75c095 cp_parser_template_declaration_after_parameters
        ../../gcc/cp/parser.c:29984
0x75c84b cp_parser_explicit_template_declaration
        ../../gcc/cp/parser.c:30250
0x75c84b cp_parser_template_declaration_after_export
        ../../gcc/cp/parser.c:30269
0x75ecf9 cp_parser_declaration
        ../../gcc/cp/parser.c:13996

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=90915,
                   |                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=92104
                 CC|                            |msebor at gcc dot gnu.org
   Last reconfirmed|                            |2020-12-17
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Martin Sebor <msebor at gcc dot gnu.org> ---
The built-in (still) isn't implemented for templates.  See also pr90915 (and
pr92104).  There's little value in raising new reports for this or in changing
the ICE to Sorry.

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |9.4

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
sorry() informs the user that the issue is known and not an "internal compiler
error" so I disagree

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Agreed, but I'd also say that it shouldn't have been accepted without the
template support, because templates are essential part of the C++ language and
so not supporting templates means C++ isn't really supported.  Design of all
the C++ FE features needs to take templates into account.

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

--- Comment #4 from Martin Sebor <msebor at gcc dot gnu.org> ---
I had stated my intent in the initial submission to complete the C++ support in
a followup patch:
https://gcc.gnu.org/pipermail/gcc-patches/2018-November/510455.html

Jason approved the initial patch with an encouragement to do so:
https://gcc.gnu.org/legacy-ml/gcc-patches/2018-10/msg01528.html

I might have gotten around to it if you and Jeff hadn't blocked for four months
(!) my fixes for a few simple ICEs in the code.  Here's the first submission of
a patch for the first ICE I found in December 2018:
https://gcc.gnu.org/pipermail/gcc-patches/2018-December/512645.html

The fixes didn't get approved and committed in r270326 until April 2019, and
many rounds of back and forth bickering over whether the built-in should accept
expressions or error out on them.

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mpolacek at gcc dot gnu.org

--- Comment #5 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
My earlier fix to emit sorry rather than ICE didn't catch this case; this patch
fixes that:

--- a/gcc/cp/parser.c
+++ b/gcc/cp/parser.c
@@ -8934,7 +8934,7 @@ cp_parser_has_attribute_expression (cp_parser *parser)
     {
       if (oper == error_mark_node)
    /* Nothing.  */;
-      else if (type_dependent_expression_p (oper))
+      else if (uses_template_parms (oper))
    sorry_at (atloc, "%<__builtin_has_attribute%> with dependent argument "
          "not supported yet");
       else

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |mpolacek at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

[Bug c++/98355] [9/10/11 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Marek Polacek <mpolacek@gcc.gnu.org>:

https://gcc.gnu.org/g:bab669f2fc643cb1673aecd177eec1c773e9e48e

commit r11-7012-gbab669f2fc643cb1673aecd177eec1c773e9e48e
Author: Marek Polacek <polacek@redhat.com>
Date:   Fri Jan 29 11:29:25 2021 -0500

    c++: Improve sorry for __builtin_has_attribute [PR98355]

    __builtin_has_attribute doesn't work in templates yet (bug 92104), so
    in r11-471 I added a sorry.  But that only caught type-dependent
    expressions and we also want to sorry on value-dependent expressions.
    This patch uses uses_template_parms, but guarded with p_t_d, because
    u_t_p sets p_t_d and then v_d_e_p considers variables with reference
    types value-dependent, which breaks builtin-has-attribute-6.c.

    This is a regression and I also plan to apply this to gcc-10.

    gcc/cp/ChangeLog:

            PR c++/98355
            * parser.c (cp_parser_has_attribute_expression): Use
            uses_template_parms instead of type_dependent_expression_p.

    gcc/testsuite/ChangeLog:

            PR c++/98355
            * g++.dg/ext/builtin-has-attribute2.C: New test.

[Bug c++/98355] [9/10 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[9/10/11 Regression] ICE in |[9/10 Regression] ICE in
                   |has_attribute, at           |has_attribute, at
                   |c-family/c-attribs.c:5628   |c-family/c-attribs.c:5628

--- Comment #7 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Fixed on trunk thus far.

[Bug c++/98355] [9/10 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Marek Polacek
<mpolacek@gcc.gnu.org>:

https://gcc.gnu.org/g:c616e86e8bc76f867242b4aba372736d1c8cbf80

commit r10-9332-gc616e86e8bc76f867242b4aba372736d1c8cbf80
Author: Marek Polacek <polacek@redhat.com>
Date:   Fri Jan 29 11:29:25 2021 -0500

    c++: Improve sorry for __builtin_has_attribute [PR98355]

    __builtin_has_attribute doesn't work in templates yet (bug 92104), so
    in r11-471 I added a sorry.  But that only caught type-dependent
    expressions and we also want to sorry on value-dependent expressions.
    This patch uses uses_template_parms, but guarded with p_t_d, because
    u_t_p sets p_t_d and then v_d_e_p considers variables with reference
    types value-dependent, which breaks builtin-has-attribute-6.c.

    This is a regression and I also plan to apply this to gcc-10.

    gcc/cp/ChangeLog:

            PR c++/98355
            * parser.c (cp_parser_has_attribute_expression): Use
            uses_template_parms instead of type_dependent_expression_p.

    gcc/testsuite/ChangeLog:

            PR c++/98355
            * g++.dg/ext/builtin-has-attribute2.C: New test.

    (cherry picked from commit bab669f2fc643cb1673aecd177eec1c773e9e48e)

[Bug c++/98355] [9 Regression] ICE in has_attribute, at c-family/c-attribs.c:5628

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98355

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
            Summary|[9/10 Regression] ICE in    |[9 Regression] ICE in
                   |has_attribute, at           |has_attribute, at
                   |c-family/c-attribs.c:5628   |c-family/c-attribs.c:5628

--- Comment #9 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Fixed in 10 too.