Otherwise, gcc will not be able to compile parts of libc++ 21 in -m32
mode, resulting in errors similar to:
In file included from /usr/src/contrib/llvm-project/libc/shared/str_to_float.h:13,
from /usr/src/contrib/llvm-project/libcxx/src/include/from_chars_floating_point.h:14,
from /usr/src/contrib/llvm-project/libcxx/src/charconv.cpp:12:
/usr/src/contrib/llvm-project/libc/src/__support/str_to_float.h: In function 'void __llvm_libc::internal::set_implicit_bit(__llvm_libc::fputil::FPBits<T>&) [with T = long double]':
/usr/src/contrib/llvm-project/libc/src/__support/str_to_float.h:77:10: error: 'struct __llvm_libc::fputil::FPBits<long double>' has no member named 'set_implicit_bit'
77 | result.set_implicit_bit(result.get_biased_exponent() != 0);
| ^~~~~~~~~~~~~~~~
PR: 292067
MFC after: 1 month
Dimitry Andric
718e5874be