Re: [PATCH] Add C++ runtime support for new 128-bit long double format

[Michael Meissner <meissner at linux dot ibm dot com>]

On Mon, Nov 12, 2018 at 11:09:45AM +0000, Jonathan Wakely wrote:
> This adds support for the new 128-bit long double format on powerpc64,
> see https://fedoraproject.org/wiki/Changes/PPC64LE_Float128_Transition
> for more details.
> 
> Most of the required changes are to the locale facets that parse and
> print long doubles, as used by iostreams for reading/writing numbers.

Thanks for the patches.  Unfortunately I find if I use the Advance Toolchain
AT12 libraries which have the <func>f128 functions, and I try to use it, I get
some redefinition errors:

In file included from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/move.h:55,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/nested_exception.h:40,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/exception:144,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/ios:39,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/ostream:38,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/iostream:39,
                 from foo.cc:1:
/home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/type_traits:340:12: error: redefinition of 'struct std::__is_floating_point_helper<long double>'
  340 |     struct __is_floating_point_helper<__float128>
      |            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/type_traits:335:12: note: previous definition of 'struct std::__is_floating_point_helper<long double>'
  335 |     struct __is_floating_point_helper<long double>
      |            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/cstdlib:77,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/ext/string_conversions.h:41,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/basic_string.h:6412,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/string:52,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/locale_classes.h:40,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/ios_base.h:41,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/ios:42,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/ostream:38,
                 from /home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/iostream:39,
                 from foo.cc:1:
/home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/std_abs.h:102:3: error: redefinition of 'constexpr long double std::abs(long double)'
  102 |   abs(__float128 __x)
      |   ^~~
/home/meissner/fsf-install-ppc64le/ieee-c++/include/c++/9.0.0/bits/std_abs.h:78:3: note: 'constexpr long double std::abs(long double)' previously defined here
   78 |   abs(long double __x)
      |   ^~~

I configured the compiler with the options:

	/home/meissner/fsf-src/ieee-c++/configure \
	--prefix=/home/meissner/fsf-install-ppc64le/ieee-c++ \
	--enable-languages=c,c++,fortran --enable-checking --enable-stage1-checking \
	--enable-gnu-indirect-function --enable-plugin --enable-decimal-float \
	--with-long-double-128 --enable-secureplt --enable-threads=posix \
	--enable-__cxa_atexit --with-cpu=power8 \
	--with-as=/opt/at12.0/bin/as --with-ld=/opt/at12.0/bin/ld \
	--with-gnu-as=/opt/at12.0/bin/as --with-gnu-ld=/opt/at12.0/bin/ld \
	--with-advance-toolchain=at12.0 --with-native-system-header-dir=/opt/at12.0/include \
	--without-ppl --without-cloog --without-isl

The test case was:

	#include <iostream>

	#ifndef TYPE
	#define TYPE long double
	#endif

	volatile TYPE a = (TYPE)3, b = (TYPE)4;

	int main (void)
	{
	  std::cout << "Value is " << a+b << "\n";
	  return 0;
	}

And I invoked the compiler with:

	$ /home/meissner/fsf-install-ppc64le/ieee-c++/bin/g++ -O2 -mabi=ieeelongdouble -Wno-psabi foo.cc

Let me know if I can help in testing future versions of the patch.

-- 
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.ibm.com, phone: +1 (978) 899-4797