Forgot to send to the ML ---------- Forwarded message --------- From: connor horman <chorma...@gmail.com> Date: Tue, Sep 16, 2025 at 15:39 Subject: Re: New OS Target: unable to emulate ‘XF’ To: H.J. Lu <hjl.to...@gmail.com>
I did this already, I think. That part is done and works (long double is 64-bit and the values of limits macros are set to what I expect for LDBL). On Tue, Sep 16, 2025 at 15:24 H.J. Lu <hjl.to...@gmail.com> wrote: > On Tue, Sep 16, 2025 at 9:26 AM connor horman via Gcc <gcc@gcc.gnu.org> > wrote: > > > > Hello. I'm currently trying to port gcc to a new OS on x86-64. When > > building libgcc I get > > In file included from ../../../libgcc/soft-fp/extendhfxf2.c:32: > > ../../../libgcc/soft-fp/extended.h:68:1: error: unable to emulate ‘XF’ > > 68 | typedef float XFtype __attribute__ ((mode (XF))); > > | ^~~~~~~ > > > > This may be because I've set `long double` as 64-bit by default, but > > __float80 should still work, so I don't know why this is a thing? Anyone > > know why I'm getting this error? > > i386.cc has: > > /* Default long double to 64-bit for 32-bit Bionic and to __float128 > for 64-bit Bionic. Also default long double to 64-bit for Intel > MCU psABI. */ > if ((TARGET_HAS_BIONIC || TARGET_IAMCU) > && !(opts_set->x_target_flags > & (MASK_LONG_DOUBLE_64 | MASK_LONG_DOUBLE_128))) > opts->x_target_flags |= (TARGET_64BIT > ? MASK_LONG_DOUBLE_128 > : MASK_LONG_DOUBLE_64); > > You need to do something similar for your OS. > > -- > H.J. >
