Re: sparc64 clang fixes

2020-04-05 Thread Mark Kettenis
> Date: Sat, 4 Apr 2020 23:46:05 +0200 (CEST) > From: Mark Kettenis > > So regress/lib/libm/msun/run-conj_test fails because clang emits > fmovqne instructions. Those instructions aren't actually implemented > and since we don't emulate them in our kernel the test gets killed > with SIGILL. > >

sparc64 clang fixes

2020-04-04 Thread Mark Kettenis
So regress/lib/libm/msun/run-conj_test fails because clang emits fmovqne instructions. Those instructions aren't actually implemented and since we don't emulate them in our kernel the test gets killed with SIGILL. The compiler isn't suppose to emit the instructions unless they are explicitly enab