[gcc r13-4117] libstdc++: Ensure std::to_chars overloads all declared in <format> [PR107720]

[gcc r13-4117] libstdc++: Ensure std::to_chars overloads all declared in <format> [PR107720]

[Jonathan Wakely]

https://gcc.gnu.org/g:f69a8299c1d95548e1539227fb7b8f5581aeb29b

commit r13-4117-gf69a8299c1d95548e1539227fb7b8f5581aeb29b
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Wed Nov 16 20:47:39 2022 +0000

    libstdc++: Ensure std::to_chars overloads all declared in <format> [PR107720]
    
    For powerpc64le we need to be able to format both of __ieee128 and
    __ibm128, so we need the std::to_chars overloads for both types to be
    visible at once. The __ieee128 overloads are always visible in C++23
    mode, because they're used to implement the _Float128 overloads. The
    __ibm128 overloads are only visible when long double is __ibm128.
    
    libstdc++-v3/ChangeLog:
    
            PR libstdc++/107720
            * include/std/format [_GLIBCXX_LONG_DOUBLE_ALT128_COMPAT]:
            Declare overloads of std::to_chars for the alternative long
            double type.

Diff:
---
 libstdc++-v3/include/std/format | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/libstdc++-v3/include/std/format b/libstdc++-v3/include/std/format
index 1afcd042bc2..f4fc85a16d2 100644
--- a/libstdc++-v3/include/std/format
+++ b/libstdc++-v3/include/std/format
@@ -1237,6 +1237,34 @@ namespace __format
   using __float128_t = __ieee128;
 # define _GLIBCXX_FORMAT_F128 1
 
+#ifdef __LONG_DOUBLE_IEEE128__
+  // These overloads exist in the library, but are not declared.
+  // Make them available as std::__format::to_chars.
+  to_chars_result
+  to_chars(char*, char*, __ibm128) noexcept
+    __asm("_ZSt8to_charsPcS_e");
+
+  to_chars_result
+  to_chars(char*, char*, __ibm128, chars_format) noexcept
+    __asm("_ZSt8to_charsPcS_eSt12chars_format");
+
+  to_chars_result
+  to_chars(char*, char*, __ibm128, chars_format, int) noexcept
+    __asm("_ZSt8to_charsPcS_eSt12chars_formati");
+#elif __cplusplus == 202002L
+  to_chars_result
+  to_chars(char*, char*, __ieee128) noexcept
+    __asm("_ZSt8to_charsPcS_u9__ieee128");
+
+  to_chars_result
+  to_chars(char*, char*, __ieee128, chars_format) noexcept
+    __asm("_ZSt8to_charsPcS_u9__ieee128St12chars_format");
+
+  to_chars_result
+  to_chars(char*, char*, __ieee128, chars_format, int) noexcept
+    __asm("_ZSt8to_charsPcS_u9__ieee128St12chars_formati");
+#endif
+
 #elif defined _GLIBCXX_LDOUBLE_IS_IEEE_BINARY128
 
   // Format 128-bit floating-point types using long double.