An updated version of patch, that takes a safer approach on not declaring a
special formatter for __float128 if its long double is possibly IEEE 128.
Please let me know if that version addressed the problem on sparc. I
thinking there are more sparc machines in compiler farm, so I am going to
check
that tomorrow.
diff --git a/libstdc++-v3/include/std/format
b/libstdc++-v3/include/std/format
index b1823db83bc..d2ac84a1709 100644
--- a/libstdc++-v3/include/std/format
+++ b/libstdc++-v3/include/std/format
@@ -2973,11 +2973,10 @@ namespace __format
};
#endif
-#if defined(__SIZEOF_FLOAT128__) && _GLIBCXX_FORMAT_F128 > 1
+#if defined(__SIZEOF_FLOAT128__) && _GLIBCXX_FORMAT_F128 == 2
// Reuse __formatter_fp<C>::format<__format::__flt128_t, Out> for
__float128.
- // This formatter is not declared if _GLIBCXX_LONG_DOUBLE_ALT128_COMPAT
is true,
- // as __float128 when present is same type as __ieee128, which may be
same as
- // long double.
+ // This formatter is only declared when __flt128_t is _Float128, as in
other
+ // cases __float128 may be same type as long double (powerpc and sparc).
template<__format::__char _CharT>
struct formatter<__float128, _CharT>
{
@@ -2995,9 +2994,6 @@ namespace __format
private:
__format::__formatter_fp<_CharT> _M_f;
-
- static_assert( !is_same_v<__float128, long double>,
- "This specialization should not be used for long
double" );
};
#endif
--
2.49.0
On Thu, May 15, 2025 at 9:20 PM Tomasz Kaminski <tkami...@redhat.com> wrote:
> Hi,
>
> I apologize for the chrum that this patch created. I have added this
> static assert to detect environments where __float128 is same as long
> double.
> From original commit message:
> >We also provide formatter<__float128, _CharT> that formats via __flt128_t.
> >As this type may be disabled (-mno-float128), extra care needs to be
> taken,
> >for situation when __float128 is same as long double. If the formatter
> would be
> >defined in such case, the formatter<long double, _CharT> would be
> generated
> >from different specializations, and have different mangling:
> > * formatter<__float128, _CharT> if __float128 is present,
> > * formatter<__format::__formattable_float, _CharT> otherwise.
> >To best of my knowledge this happens only on ppc64 for __ieee128 and
> __float128,
> >so the formatter is not defined in this case. static_assert is added to
> detect
> >other configurations like that. In such case we should replace it with
> constra
>
> I do not have access to a machine with that setup to confirm,
> but could you please check if the following patch fixes this.
> ---
> libstdc++-v3/include/std/format | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/libstdc++-v3/include/std/format
> b/libstdc++-v3/include/std/format
> index b1823db83bc..a870aa3f3ea 100644
> --- a/libstdc++-v3/include/std/format
> +++ b/libstdc++-v3/include/std/format
> @@ -2979,6 +2979,7 @@ namespace __format
> // as __float128 when present is same type as __ieee128, which may be
> same as
> // long double.
> template<__format::__char _CharT>
> + requires (!is_same_v<__float128, long double>)
> struct formatter<__float128, _CharT>
> {
> formatter() = default;
> @@ -2995,9 +2996,6 @@ namespace __format
>
> private:
> __format::__formatter_fp<_CharT> _M_f;
> -
> - static_assert( !is_same_v<__float128, long double>,
> - "This specialization should not be used for long
> double" );
> };
> #endif
>
> --
> 2.49.0
>
>
>
> On Thu, May 15, 2025 at 7:46 PM Rainer Orth <r...@cebitec.uni-bielefeld.de>
> wrote:
>
>> Hi Jonathan,
>>
>> > On Thu, 15 May 2025 at 15:02, Rainer Orth <r...@cebitec.uni-bielefeld.de>
>> wrote:
>> >>
>> >> Hi Jonathan,
>> >>
>> >> >> > this patch broke Solaris bootstrap, both i386-pc-solaris2.11 and
>> >> >> > sparc-sun-solaris2.11:
>> >> >> >
>> >> >> > In file included from
>> >> >> > /vol/gcc/src/hg/master/local/libstdc++-v3/src/c++20/format.cc:29:
>> >> >> >
>> /var/gcc/regression/master/11.4-gcc/build/i386-pc-solaris2.11/libstdc++-v3/include/format:
>> >> >> > In member function ‘typename std::basic_format_context<_Out,
>> >> >> > _CharT>::iterator std::formatter<__float128,
>> >> >> > _CharT>::format(__float128, std::basic_format_context<_Out,
>> _CharT>&)
>> >> >> > const’:
>> >> >> >
>> /var/gcc/regression/master/11.4-gcc/build/i386-pc-solaris2.11/libstdc++-v3/include/format:2994:41:
>> >> >> > error: ‘__flt128_t’ is not a member of ‘std::__format’; did you
>> mean
>> >> >> > ‘__bflt16_t’? [-Wtemplate-body]
>> >> >> > 2994 | { return _M_f.format((__format::__flt128_t)__u,
>> __fc); }
>> >> >> > | ^~~~~~~~~~
>> >> >> > | __bflt16_t
>> >> >> >
>> >> >> > and one more instance.
>> >> >>
>> >> >> And on x86_64-darwin too.
>> >> >
>> >> > Tomasz, should this be:
>> >> >
>> >> > --- a/libstdc++-v3/include/std/format
>> >> > +++ b/libstdc++-v3/include/std/format
>> >> > @@ -2973,7 +2973,7 @@ namespace __format
>> >> > };
>> >> > #endif
>> >> >
>> >> > -#if defined(__SIZEOF_FLOAT128__) && _GLIBCXX_FORMAT_F128 != 1
>> >> > +#if defined(__SIZEOF_FLOAT128__) && _GLIBCXX_FORMAT_F128 > 1
>> >> > // Reuse __formatter_fp<C>::format<__format::__flt128_t, Out> for
>> __float128.
>> >> > // This formatter is not declared if
>> >> > _GLIBCXX_LONG_DOUBLE_ALT128_COMPAT is true,
>> >> > // as __float128 when present is same type as __ieee128, which may
>> be same as
>> >> >
>> >>
>> >> with this patch applied, I could link libstdc++.so. I'll run a full
>> >> bootstrap later today.
>> >
>> >
>> > Good to know, thanks. Tomasz already pushed that change as
>> > r16-647-gd010a39b9e788a
>> > so trunk should be OK now.
>>
>> it is on Solaris/i386, but sparc is broken in a different way now:
>>
>> /var/gcc/regression/master/11.4-gcc/build/sparc-sun-solaris2.11/libstdc++-v3/include/format:2999:23:
>> error: static assertion failed: This specialization should not be used for
>> long double
>> 2999 | static_assert( !is_same_v<__float128, long double>,
>> | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> /var/gcc/regression/master/11.4-gcc/build/sparc-sun-solaris2.11/libstdc++-v3/include/format:2999:23:
>> note: ‘!(bool)std::is_same_v<long double, long double>’ evaluates to false
>> /var/gcc/regression/master/11.4-gcc/build/sparc-sun-solaris2.11/libstdc++-v3/include/format:
>> In instantiation of ‘struct std::formatter<long double, char>’:
>> /var/gcc/regression/master/11.4-gcc/build/sparc-sun-solaris2.11/libstdc++-v3/include/type_traits:3559:54:
>> required from ‘constexpr const bool
>> std::is_default_constructible_v<std::formatter<long double, char> >’
>> 3559 | inline constexpr bool is_default_constructible_v =
>> __is_constructible(_Tp);
>> |
>> ^~~~~~~~~~~~~~~~~~~~~~~
>> [...]
>>
>> I've a local patch in tree to support __float128 on SPARC, so I'll try
>> with an unmodified tree first. However, 2 days ago I could bootstrap
>> with that included just fine.
>>
>> Rainer
>>
>> --
>>
>> -----------------------------------------------------------------------------
>> Rainer Orth, Center for Biotechnology, Bielefeld University
>>
>>
