[PATCH] newlib: Disable ld128 math library build for aarch64 windows

[Corinna Vinschen]

On Feb 24 08:56, Thirumalai Nagalingam wrote:
> Hello,
> On AArch64 Windows targets (Cygwin and MinGW), long double has the same
> representation & precision as double. The _fpmath.h header and the
> ld128 libm implementation assume extended precision long double
> semantics, which is not valid for these targets.
> 
> This patch detects AArch64 Windows targets and disables HAVE_FPMATH_H
> accordingly, preventing inclusion of incompatible ld128 math routines.
> This fixes build & ABI issues in libm for Cygwin AArch64.
> 
> Patch is attached & In-lined for reference.
> 
> Thanks & regards,
> Thirumalai N
> In-lined patch:
> ---
>  newlib/configure         |  8 ++++++++
>  newlib/libc/acinclude.m4 | 11 ++++++++++-
>  newlib/libm/Makefile.inc |  2 ++
>  3 files changed, 20 insertions(+), 1 deletion(-)

Pushed.

Thanks,
Corinna