Add --disable-tls to GCC?

Eric Gallager egall at gwmail.gwu.edu
Sat Nov 13 16:57:28 UTC 2021


On Wed, Aug 2, 2017 at 12:24 PM Marcus Calhoun-Lopez
<mcalhoun at macports.org> wrote:
>
> The file ${prefix}/include/gcc6/c++/${build_arch}-apple-darwin${os.major}/bits/c++config.h
> has different values for _GLIBCXX_HAVE_TLS on different platforms.
> For ${os.major}  <  15, _GLIBCXX_HAVE_TLS is defined as 1.
> For ${os.major} >= 15, _GLIBCXX_HAVE_TLS is undefined.
> I am not sure why the availability of thread-local storage changed,
> but for consistency, should --disable-tls be added to the GCC
> configure script (see https://gcc.gnu.org/install/configure.html)?
>
> The reason I bring this up is that the following code works macOS
> Sierra and OS X El Capitan:
> ------------------------------------------------------------------------------------------------------------------------------
>     clang++-mp-4.0 -std=c++11 -stdlib=macports-libstdc++ future.cxx
> ------------------------------------------------------------------------------------------------------------------------------
>    #include <iostream>
>    #include <future>
>
>   int main()
>   {
>       std::future<int> f1 = std::async([](){ return 42; });
>       f1.wait();
>       std::cout << "Magic number is: " << f1.get() << std::endl;
>   }
> ------------------------------------------------------------------------------------------------------------------------------
>
> It does *not* work for OS versions prior to El Capitan with the error:
> ------------------------------------------------------------------------------------------------------------------------------
> Undefined symbols for architecture x86_64:
>   "std::__once_call", referenced from:
>       void std::call_once<void
> (std::__future_base::_State_baseV2::*)(std::function<std::unique_ptr<std::__future_base::_Result_base,
> std::__future_base::_Result_base::_Deleter> ()>*, bool*),
> std::__future_base::_State_baseV2*,
> std::function<std::unique_ptr<std::__future_base::_Result_base,
> std::__future_base::_Result_base::_Deleter> ()>*,
> bool*>(std::once_flag&, void
> (std::__future_base::_State_baseV2::*&&)(std::function<std::unique_ptr<std::__future_base::_Result_base,
> std::__future_base::_Result_base::_Deleter> ()>*, bool*),
> std::__future_base::_State_baseV2*&&,
> std::function<std::unique_ptr<std::__future_base::_Result_base,
> std::__future_base::_Result_base::_Deleter> ()>*&&, bool*&&) in
> future-b039c9.o
> ------------------------------------------------------------------------------------------------------------------------------
>   clang++-mp-4.0 -std=c++11 -stdlib=macports-libstdc++ -femulated-tls future.cxx
> ------------------------------------------------------------------------------------------------------------------------------
> compiles fine but experiences a segmentation fault at runtime.
>
> When I build GCC with --disable-tls, everything works on all platforms.
>
> I realize that -stdlib=macports-libstdc++ is a highly unsupported
> configuration, but --disable-tls would not seem to affect newer OSs.
>
> Does anyone have any insight into this?
> Does anyone have any objections to adding --disable-tls?
>
> Thanks,
> Marcus

Hi, could you open a bug upstream with GCC about this? One of the
upstream GCC devs is wondering why MacPorts needs this configure flag,
so if they had a bug report to work with, they could probably fix
whatever is requiring MacPorts to use that flag.


More information about the macports-dev mailing list