[Bug bootstrap/100186] New: lto-wrapper.c assumes std::thread

[Bug bootstrap/100186] New: lto-wrapper.c assumes std::thread

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

            Bug ID: 100186
           Summary: lto-wrapper.c assumes std::thread
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dje at gcc dot gnu.org
  Target Milestone: ---
            Target: powerpc-ibm-aix*

lto-wrapper.c:1288:37: error 'std::thread' has not been declared

unsigned long nthreads_var = std::thread::hardware_concurrency ();

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0
   Last reconfirmed|                            |2021-04-21
                 CC|                            |mliska at suse dot cz
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from David Edelsohn <dje at gcc dot gnu.org> ---
Confirmed.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #2 from David Edelsohn <dje at gcc dot gnu.org> ---
Created attachment 50650
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50650&action=edit
pre-processed source for lto-wrapper.c

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

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> ---
Yeah, only GCC 10 and later seems to define std::thread in <thread>
unconditionally, older releases made it conditional on #ifdef
_GLIBCXX_HAS_GTHREADS or on #if defined(_GLIBCXX_HAS_GTHREADS) &&
defined(_GLIBCXX_USE_C99_STDINT_TR1)

The fact that something is in C++11 doesn't mean we can assume it in GCC,
while GCC 4.8.x which we still want to support had roughly complete C++11 FE
support, I think the libstdc++ side wasn't there yet.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
And opinions differ on whether "define it but make it useless" or "don't define
it" is the right answer.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #5 from David Edelsohn <dje at gcc dot gnu.org> ---
For example, in GCC 6:

#if defined(_GLIBCXX_HAS_GTHREADS) && defined(_GLIBCXX_USE_C99_STDINT_TR1)

namespace std _GLIBCXX_VISIBILITY(default)
{
_GLIBCXX_BEGIN_NAMESPACE_VERSION

  /**
   * @defgroup threads Threads
   * @ingroup concurrency
   *
   * Classes for thread support.
   * @{
   */

  /// thread
  class thread
  {

If we are going to support bootstrap with releases of GCC back to GCC 6 or GCC
4.8, the source code cannot assume that std::thread exists.

However, it's probably reasonable to assume hardware_concurrency of 1 when
compiling in a mode without GTHREADS.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #6 from Martin Liška <marxin at gcc dot gnu.org> ---
Mine.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #7 from Martin Liška <marxin at gcc dot gnu.org> ---
> The fact that something is in C++11 doesn't mean we can assume it in GCC,
> while GCC 4.8.x which we still want to support had roughly complete C++11 FE
> support, I think the libstdc++ side wasn't there yet.

Do we really support it? https://gcc.gnu.org/install/prerequisites.html
mentions that an 'ISO C++11 compiler' is required.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Martin Liška from comment #7)
> > The fact that something is in C++11 doesn't mean we can assume it in GCC,
> > while GCC 4.8.x which we still want to support had roughly complete C++11 FE
> > support, I think the libstdc++ side wasn't there yet.
> 
> Do we really support it? https://gcc.gnu.org/install/prerequisites.html
> mentions that an 'ISO C++11 compiler' is required.

Yes, we do, support back to 4.8.x was part of the switch to C++11 language
decision.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

Martin Liška <marxin at gcc dot gnu.org> changed:

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

--- Comment #9 from Martin Liška <marxin at gcc dot gnu.org> ---
I reverted the commit in g:0a18305ee11e139838771f96c5a037a29606236e.

[Bug bootstrap/100186] lto-wrapper.c assumes std::thread

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100186

--- Comment #10 from Martin Liška <marxin at gcc dot gnu.org> ---
> Yes, we do, support back to 4.8.x was part of the switch to C++11 language
> decision.

Can you please document that requirement?
I really don't get it, we decided to move to a more recent C++ standard, but we
still support a compiler with partial (or incomplete) support of the C++
standard.