[Bug c++/94162] New: ICE [neg] bad return type in defaulted <=>

[Bug c++/94162] New: ICE [neg] bad return type in defaulted <=>

[dacamara.cameron at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

            Bug ID: 94162
           Summary: ICE [neg] bad return type in defaulted <=>
           Product: gcc
           Version: 10.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dacamara.cameron at gmail dot com
  Target Milestone: ---

Whenever the return type of a defaulted <=> is not a comparison category gcc
ICEs:

#include <compare>

struct S {
  int a;
  bool operator<=>(const S&) const = default;
};

bool b = S{} < S{};

$ gcc -std=c++2a -c t.cpp

t.cpp:5:8: internal compiler error: in genericize_spaceship, at
cp/method.c:1050

    5 |   bool operator<=>(const S&) const = default;

      |        ^~~~~~~~

Please submit a full bug report,

with preprocessed source if appropriate.

See <https://gcc.gnu.org/bugs/> for instructions.

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
It isn't clear to me what exactly disallows it, perhaps
http://eel.is/c++draft/class.spaceship#2.2
?
For auto return type
http://eel.is/c++draft/class.spaceship#4
defines what return type it should have.
If the explicit return type isn't auto, but is one of the
std::{strong,weak,partial}_ordering, we don't ICE and accept it, should we and
what behavior should it have?
#include <compare>
struct S {
  float a;
  std::strong_ordering operator<=>(const S&) const = default;
};
bool b = S{} < S{};
struct T {
  std::partial_ordering operator<=>(const T&) const = default;
};
bool c = T{} < T{};
For S, the auto return type would be std::partial_ordering and in the generated
body we just assume the floats will not be unordered.
So, for bool, shall it be accepted and handled some way, or shall it be
deleted, or result in immediate error (ill-formed)?
What about even weirder types (say float or int * or some arbitrary class)?

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[dacamara.cameron at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #2 from Cameron <dacamara.cameron at gmail dot com> ---
(In reply to Jakub Jelinek from comment #1)
> It isn't clear to me what exactly disallows it, perhaps
> http://eel.is/c++draft/class.spaceship#2.2
> ?
> For auto return type
> http://eel.is/c++draft/class.spaceship#4
> defines what return type it should have.
> If the explicit return type isn't auto, but is one of the
> std::{strong,weak,partial}_ordering, we don't ICE and accept it, should we
> and what behavior should it have?
> #include <compare>
> struct S {
>   float a;
>   std::strong_ordering operator<=>(const S&) const = default;
> };
> bool b = S{} < S{};
> struct T {
>   std::partial_ordering operator<=>(const T&) const = default;
> };
> bool c = T{} < T{};
> For S, the auto return type would be std::partial_ordering and in the
> generated body we just assume the floats will not be unordered.
> So, for bool, shall it be accepted and handled some way, or shall it be
> deleted, or result in immediate error (ill-formed)?
> What about even weirder types (say float or int * or some arbitrary class)?

As you point out, the standard isn't clear about what to do in the case where
the comparison function does not return 'auto' or any of the comparison type
forms.

Our compiler (MSVC) now opts to issue a diag at the point of definition when a
nonsense return type is specified.  There is, unfortunately, room for
implementation divergence here.

Not ICEing--as we once did--is a good start though :).

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |poul250 at yandex dot ru

--- Comment #3 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
*** Bug 96627 has been marked as a duplicate of this bug. ***

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nunoplopes at sapo dot pt

--- Comment #4 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
*** Bug 96060 has been marked as a duplicate of this bug. ***

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

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> ---
Test from bug 96060:

#include <set>

struct xx {
    int a;
    int operator<=>(const xx&) const = default;
};

int f(std::set<xx> &x) {
    x.emplace(0);
    return 0;
}

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2020-08-18

--- Comment #6 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Test from Bug 96627:

#include <iostream>

struct Float {
    float value;

    constexpr Float(float value) : value (value) {}

    constexpr bool operator == (Float other) const noexcept {
        return abs(value - other.value) < 0.5;
    }

    constexpr bool operator != (Float other) const noexcept {
        return !(*this == other);
    }

    constexpr bool operator <=> (const Float& other) const noexcept = default;
};

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |msharov at users dot sourceforge.n
                   |                            |et

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
*** Bug 99429 has been marked as a duplicate of this bug. ***

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Test from PR99429:
namespace std {
struct strong_ordering {
  int _v;
  constexpr strong_ordering (int v) :_v(v) {}
  constexpr operator int (void) const { return _v; }
  static const strong_ordering less;
  static const strong_ordering equal;
  static const strong_ordering greater;
};
constexpr strong_ordering strong_ordering::less = -1;
constexpr strong_ordering strong_ordering::equal = 0;
constexpr strong_ordering strong_ordering::greater = 1;
}

template <unsigned long N>
struct duration {
  static constexpr const long period = N;
  constexpr duration (void) = default;
  constexpr duration (const duration& d) = default;
  constexpr bool operator== (const duration& d) const = default;
  constexpr bool operator<=> (const duration& d) const = default;
  long _d;
};

using nanoseconds = duration<1>;
using microseconds = duration<nanoseconds::period * 1000>;

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |11.1.0
      Known to fail|                            |10.1.0

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
None of the testcases ICE in GCC 11+

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[dacamara.cameron at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #10 from Cameron <dacamara.cameron at gmail dot com> ---
(In reply to Andrew Pinski from comment #9)
> None of the testcases ICE in GCC 11+

I notice that if I change the repro to:

#include <compare>

struct S {
  int a;
  int* operator<=>(const S&) const = default;
};

bool b = S{} < S{};


Notice the use of 'int*' as the comparison type.  GCC11 series will still
crash.

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|11.1.0                      |

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Cameron from comment #10)
> Notice the use of 'int*' as the comparison type.  GCC11 series will still
> crash.

Confirmed, this testcase ICEs on the trunk also.

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
At least the #c8 testcase got fixed with
r12-619-gfc178519771db508c03611cff4a1466cf67fce1d
but that hasn't been backported to 11.x.

The ICE with int * return type is because POINTER_TYPE has NULL
TYPE_LINKAGE_IDENTIFIER, and cat_type_for is called on such type.
--- gcc/cp/method.c.jj  2021-06-28 22:38:27.014617793 +0200
+++ gcc/cp/method.c     2021-08-09 10:59:02.324443669 +0200
@@ -1029,6 +1029,8 @@ is_cat (tree type, comp_cat_tag tag)
 static comp_cat_tag
 cat_tag_for (tree type)
 {
+  if (!TYPE_LINKAGE_IDENTIFIER (type))
+    return cc_last;
   for (int i = 0; i < cc_last; ++i)
     {
       comp_cat_tag tag = (comp_cat_tag)i;
would fix this (or shall we instead test if (!CLASS_TYPE_P (type)) or something
similar?

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

--- Comment #13 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:9b7ab853bf33106fd0539e36d6ce7730269026e1

commit r12-2873-g9b7ab853bf33106fd0539e36d6ce7730269026e1
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Aug 12 09:16:13 2021 +0200

    c++: Fix ICE on defaulted spaceship with pointer return type [PR94162]

    The spaceship-synth-neg6.C testcase ICEs because we call cat_tag_for
    on the explicit return type, but pointer types don't have
    TYPE_LINKAGE_IDENTIFIER.  The patch fixes that by checking for
    CLASS_TYPE_P only and also adds verification that it is in std
    namespace, so we don't return non-cc_last for
my_namespace::partial_ordering.

    The g++.dg/cpp2a/spaceship-synth11.C testcase is from a PR that has been
    fixed with r12-619-gfc178519771db508c03611cff4a1466cf67fce1d (but
    not backported to 11).

    2021-08-12  Jakub Jelinek  <jakub@redhat.com>

    gcc/cp/
            PR c++/94162
            * method.c (cat_tag_for): Return cc_last for !CLASS_TYPE_P
            or for classes not in std namespace.
    gcc/testsuite/
            PR c++/99429
            * g++.dg/cpp2a/spaceship-synth11.C: New test.

            PR c++/94162
            * g++.dg/cpp2a/spaceship-synth-neg6.C: New test.

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |asolokha at gmx dot com

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 101943 has been marked as a duplicate of this bug. ***

[Bug c++/94162] ICE [neg] bad return type in defaulted <=>

[arthur.j.odwyer at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94162

Arthur O'Dwyer <arthur.j.odwyer at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |arthur.j.odwyer at gmail dot com

--- Comment #15 from Arthur O'Dwyer <arthur.j.odwyer at gmail dot com> ---
The test case in #c10 seems to be fixed since GCC 12; the rest were fixed since
GCC 11. Should this bug be RESOLVED FIXED at this point?
https://godbolt.org/z/d16x181xh