Noticed bootstrap failure on r14-3320-g7f8d730a563983 today: $ ~/dev/git/gcc/configure --build=i686-unknown-linux-gnu --host=i686-unknown-linux-gnu --target=i686-unknown-linux-gnu && make ... /tmp/gb/./gcc/xgcc -shared-libgcc -B/tmp/gb/./gcc -nostdinc++ -L/tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/src -L/tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/src/.libs -L/tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/libsupc++/.libs -B/usr/local/i686-unknown-linux-gnu/bin/ -B/usr/local/i686-unknown-linux-gnu/lib/ -isystem /usr/local/i686-unknown-linux-gnu/include -isystem /usr/local/i686-unknown-linux-gnu/sys-include -fno-checking -x c++-header -nostdinc++ -g -O2 -D_GNU_SOURCE -I/tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/include/i686-unknown-linux-gnu -I/tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/include -I/home/slyfox/dev/git/gcc/libstdc++-v3/libsupc++ -O2 -g -std=gnu++0x /home/slyfox/dev/git/gcc/libstdc++-v3/include/precompiled/stdc++.h \ -o i686-unknown-linux-gnu/bits/stdc++.h.gch/O2ggnu++0x.gch In file included from /home/slyfox/dev/git/gcc/libstdc++-v3/include/precompiled/stdc++.h:55: /tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/include/limits: In static member function 'static constexpr _Float16 std::numeric_limits<_Float16>::min()': /tmp/gb/i686-unknown-linux-gnu/libstdc++-v3/include/limits:1986:1: error: SSE register return with SSE2 disabled
Ah, looks like the 16-bit types still need to be enabled conditionally.
Created attachment 55756 [details] Revert pre-C++23 support for 16-bit float types I'll test this patch as soon as I can get an i686 system set up.
The patch fixed gcc buid on i686-linux for me. Thank you!
The master branch has been updated by Jonathan Wakely <redi@gcc.gnu.org>: https://gcc.gnu.org/g:27d0cfcb2b33de07351d9cb26f62d04d7416716c commit r14-3329-g27d0cfcb2b33de07351d9cb26f62d04d7416716c Author: Jonathan Wakely <jwakely@redhat.com> Date: Fri Aug 18 13:10:15 2023 +0100 libstdc++: Revert pre-C++23 support for 16-bit float types [PR111060] In r14-3304-g1a566fddea212a and r14-3305-g6cf214b4fc97f5 I tried to enable std::format for 16-bit float types before C++23. This causes errors for targets where the types are defined but can't actually be used, e.g. i686 without sse2. Make the std::numeric_limits and std::formatter specializations for _Float16 and __bfloat16_t depend on the __STDCPP_FLOAT16_T__ and __STDCPP_BFLOAT16_T__ macros again, so they're only defined for C++23 when the type is fully supported. This is OK because the main point of my earlier commits was to add better support for _Float32 and _Float64. It seems fine for the new 16-bit types to only be supported for C++23, as they were never present before GCC 13 anyway. libstdc++-v3/ChangeLog: PR target/111060 * include/std/format (formatter): Only define specializations for 16-bit floating-point types for C++23. * include/std/limits (numeric_limits): Likewise.
Fixed, thanks for the report.