[Bug c++/95822] New: [coroutines] compiler internal error with local object with noexcept false destructor

[Bug c++/95822] New: [coroutines] compiler internal error with local object with noexcept false destructor

[victor.burckel at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

            Bug ID: 95822
           Summary: [coroutines] compiler internal error with local object
                    with noexcept false destructor
           Product: gcc
           Version: 10.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: victor.burckel at gmail dot com
  Target Milestone: ---

It seems that when a coroutine contains a local object whose destructor is
marked as noexcept(false), gcc reaches an unexpected state and aborts.
When compiling the following code with gcc10.1.0 (-std=c++20 -fcoroutines), I
get a compiler internal error: 
> internal compiler error: in gimplify_var_or_parm_decl, at gimplify.c:2830

Making the destructor noexcept fixes the error. I  got the error when using
boost log within a couroutine. It uses a record_pump whose destructor is not
noexcept.

I managed to reproduce with godbolt (both gcc10.1 and gcc trunk give me the
assertion)
https://godbolt.org/z/6JtKsU

#include <coroutine>

struct task {
  struct promise_type {
    auto initial_suspend() noexcept { return std::suspend_always{}; }
    auto final_suspend() noexcept { return std::suspend_always{}; }
    void return_void() {}
    task get_return_object() { return task{}; }
    void unhandled_exception() noexcept {}
  };

  ~task() noexcept {}

  bool await_ready() const noexcept { return false; }
  void await_suspend(std::coroutine_handle<>) noexcept {}
  void await_resume() noexcept {}
};

struct Error {
  ~Error() noexcept(false) {}
};

task g();
task f() {
  Error error;
  co_await g();
}

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

Iain Sandoe <iains at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-06-23
   Target Milestone|---                         |10.2
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |iains at gcc dot gnu.org
           Keywords|                            |ice-on-valid-code
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Iain Sandoe <iains at gcc dot gnu.org> ---
thanks for the report

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|10.2                        |10.3

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 10.2 is released, adjusting target milestone.

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

--- Comment #3 from Iain Sandoe <iains at gcc dot gnu.org> ---
Created attachment 50196
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50196&action=edit
Patch for testing

this needs some wider testing.

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Iain D Sandoe <iains@gcc.gnu.org>:

https://gcc.gnu.org/g:7005a50d0121954031a223ea5a6c57aaa7e3efd3

commit r11-7402-g7005a50d0121954031a223ea5a6c57aaa7e3efd3
Author: Iain Sandoe <iain@sandoe.co.uk>
Date:   Tue Feb 16 09:07:32 2021 +0000

    coroutines : Remove throwing_cleanup marks from the ramp [PR95822].

    The FE contains a mechanism for cleaning up return expressions if a
    function throws during the execution of cleanups prior to the return.

    If the original function has a return value with a non-trivial DTOR
    and the body contains a var with a DTOR that might throw, the function
    decl is marked "throwing_cleanup".

    However, we do not [in the coroutine ramp function, which is
    synthesised], use any body var types with DTORs that might throw.

    The original body [which will then contain the type with the throwing
    DTOR] is transformed into the actor function which only contains void
    returns, and is also wrapped in a try-catch block.

    So (a) the 'throwing_cleanup' is no longer correct for the ramp and
       (b) we do not need to transfer it to the actor which only contains
           void returns.

    gcc/cp/ChangeLog:

            PR c++/95822
            * coroutines.cc (morph_fn_to_coro): Unconditionally remove any
            set throwing_cleanup marker.

    gcc/testsuite/ChangeLog:

            PR c++/95822
            * g++.dg/coroutines/pr95822.C: New test.

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Iain D Sandoe
<iains@gcc.gnu.org>:

https://gcc.gnu.org/g:9f13c7ad960a0f2e150dd2f668a8d155b76389db

commit r10-9510-g9f13c7ad960a0f2e150dd2f668a8d155b76389db
Author: Iain Sandoe <iain@sandoe.co.uk>
Date:   Tue Feb 16 09:07:32 2021 +0000

    coroutines : Remove throwing_cleanup marks from the ramp [PR95822].

    The FE contains a mechanism for cleaning up return expressions if a
    function throws during the execution of cleanups prior to the return.

    If the original function has a return value with a non-trivial DTOR
    and the body contains a var with a DTOR that might throw, the function
    decl is marked "throwing_cleanup".

    However, we do not [in the coroutine ramp function, which is
    synthesised], use any body var types with DTORs that might throw.

    The original body [which will then contain the type with the throwing
    DTOR] is transformed into the actor function which only contains void
    returns, and is also wrapped in a try-catch block.

    So (a) the 'throwing_cleanup' is no longer correct for the ramp and
       (b) we do not need to transfer it to the actor which only contains
           void returns.

    gcc/cp/ChangeLog:

            PR c++/95822
            * coroutines.cc (morph_fn_to_coro): Unconditionally remove any
            set throwing_cleanup marker.

    gcc/testsuite/ChangeLog:

            PR c++/95822
            * g++.dg/coroutines/pr95822.C: New test.

    (cherry picked from commit 7005a50d0121954031a223ea5a6c57aaa7e3efd3)

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

Iain Sandoe <iains at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #6 from Iain Sandoe <iains at gcc dot gnu.org> ---
fixed for master and 10.3

[Bug c++/95822] [coroutines] compiler internal error with local object with noexcept false destructor

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95822

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |familiebaumanns at gmail dot com

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 96141 has been marked as a duplicate of this bug. ***