[libcxx] r314949 - [libc++] Allow users to explicitly specify ABI

Wed Oct 4 18:10:51 PDT 2017

On Wed, Oct 4, 2017 at 5:44 PM, Shoaib Meenai via cfe-commits <
cfe-commits at lists.llvm.org> wrote:

> Author: smeenai
> Date: Wed Oct  4 16:44:38 2017
> New Revision: 314949
>
> URL: http://llvm.org/viewvc/llvm-project?rev=314949&view=rev
> Log:
> [libc++] Allow users to explicitly specify ABI
>
> libc++'s current heuristic for detecting Itanium vs. Microsoft ABI falls
> short in some cases. For example, it will detect windows-itanium targets
> as using the Microsoft ABI, since they set `_MSC_VER` (for compatibility
> with Microsoft headers). Leave the current heuristic in place by default
> but also allow users to explicitly specify the ABI if need be.
>
> Modified:
>     libcxx/trunk/CMakeLists.txt
>     libcxx/trunk/include/__config
>     libcxx/trunk/include/__config_site.in
>
> Modified: libcxx/trunk/CMakeLists.txt
> URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/
> CMakeLists.txt?rev=314949&r1=314948&r2=314949&view=diff
> ============================================================
> ==================
> --- libcxx/trunk/CMakeLists.txt (original)
> +++ libcxx/trunk/CMakeLists.txt Wed Oct  4 16:44:38 2017
> @@ -99,6 +99,8 @@ cmake_dependent_option(LIBCXX_INSTALL_EX
>          "LIBCXX_ENABLE_EXPERIMENTAL_LIBRARY;LIBCXX_INSTALL_LIBRARY" OFF)
>  set(LIBCXX_ABI_VERSION 1 CACHE STRING "ABI version of libc++.")
>  option(LIBCXX_ABI_UNSTABLE "Unstable ABI of libc++." OFF)
> +option(LIBCXX_ABI_ITANIUM "Ignore auto-detection and force use of the
> Itanium ABI.")
> +option(LIBCXX_ABI_MICROSOFT "Ignore auto-detection and force use of the
> Microsoft ABI.")
>

Shouldn't these specify a default option?

>  option(LIBCXX_USE_COMPILER_RT "Use compiler-rt instead of libgcc" OFF)
>
>  if (NOT LIBCXX_ENABLE_SHARED AND NOT LIBCXX_ENABLE_STATIC)
> @@ -337,6 +339,10 @@ if (LIBCXX_HAS_MUSL_LIBC AND NOT LIBCXX_
>                        "when building for Musl with LIBCXX_HAS_MUSL_LIBC.")
>  endif()
>
> +if (LIBCXX_ABI_ITANIUM AND LIBCXX_ABI_MICROSOFT)
> +  message(FATAL_ERROR "Only one of LIBCXX_ABI_ITANIUM and
> LIBCXX_ABI_MICROSOFT can be specified.")
> +endif ()
> +
>  #===========================================================
> ====================
>  # Configure System
>  #===========================================================
> ====================
> @@ -594,6 +600,8 @@ if (NOT LIBCXX_ABI_VERSION EQUAL "1")
>    config_define(${LIBCXX_ABI_VERSION} _LIBCPP_ABI_VERSION)
>  endif()
>  config_define_if(LIBCXX_ABI_UNSTABLE _LIBCPP_ABI_UNSTABLE)
> +config_define_if(LIBCXX_ABI_ITANIUM _LIBCPP_ABI_ITANIUM)
> +config_define_if(LIBCXX_ABI_MICROSOFT _LIBCPP_ABI_MICROSOFT)
>
> I'm not a fan of the direction this is going in. It seems to require the
generation and use of a __config_site header
in all cases where it's used. I don't think we want to require that in the
common case where you're using Itanium on
Linux or Windows.

However, like you said, the attempt of this is to override the automatic
detection done in the headers. Maybe the name
should reflect that (Ex. _LIBCPP_ABI_ITANIUM_OVERRIDE)? And then the
autodetection can operate by checking for an override first?

>  config_define_if_not(LIBCXX_ENABLE_GLOBAL_FILESYSTEM_NAMESPACE
> _LIBCPP_HAS_NO_GLOBAL_FILESYSTEM_NAMESPACE)
>  config_define_if_not(LIBCXX_ENABLE_STDIN _LIBCPP_HAS_NO_STDIN)
>
> Modified: libcxx/trunk/include/__config
> URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/include/_
> _config?rev=314949&r1=314948&r2=314949&view=diff
> ============================================================
> ==================
> --- libcxx/trunk/include/__config (original)
> +++ libcxx/trunk/include/__config Wed Oct  4 16:44:38 2017
> @@ -157,11 +157,15 @@
>
>  // FIXME: ABI detection should be done via compiler builtin macros. This
>  // is just a placeholder until Clang implements such macros. For now
> assume
> -// that Windows compilers pretending to be MSVC++ target the microsoft
> ABI.
> -#if defined(_WIN32) && defined(_MSC_VER)
> -# define _LIBCPP_ABI_MICROSOFT
> -#else
> -# define _LIBCPP_ABI_ITANIUM
> +// that Windows compilers pretending to be MSVC++ target the Microsoft
> ABI,
> +// and allow the user to explicitly specify the ABI to handle cases where
> this
> +// heuristic falls short.
> +#if !defined(_LIBCPP_ABI_ITANIUM) && !defined(_LIBCPP_ABI_MICROSOFT)
> +# if defined(_WIN32) && defined(_MSC_VER)
> +#  define _LIBCPP_ABI_MICROSOFT
> +# else
> +#  define _LIBCPP_ABI_ITANIUM
> +# endif
>  #endif
>
>  // Need to detect which libc we're using if we're on Linux.
>
> Modified: libcxx/trunk/include/__config_site.in
> URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/include/_
> _config_site.in?rev=314949&r1=314948&r2=314949&view=diff
> ============================================================
> ==================
> --- libcxx/trunk/include/__config_site.in (original)
> +++ libcxx/trunk/include/__config_site.in Wed Oct  4 16:44:38 2017
> @@ -12,6 +12,8 @@
>
>  #cmakedefine _LIBCPP_ABI_VERSION @_LIBCPP_ABI_VERSION@
>  #cmakedefine _LIBCPP_ABI_UNSTABLE
> +#cmakedefine _LIBCPP_ABI_ITANIUM
> +#cmakedefine _LIBCPP_ABI_MICROSOFT
>  #cmakedefine _LIBCPP_HAS_NO_GLOBAL_FILESYSTEM_NAMESPACE
>  #cmakedefine _LIBCPP_HAS_NO_STDIN
>  #cmakedefine _LIBCPP_HAS_NO_STDOUT
>
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20171004/3a1af395/attachment-0001.html>