On Sat, 4 Mar 2023 12:35:24 +0800 David Gow <david...@google.com> wrote:
>
> [-- Attachment #1.1: Type: text/plain, Size: 2159 bytes --]
>
> On Fri, 3 Mar 2023 at 16:37, Roberto Sassu
> <roberto.sa...@huaweicloud.com> wrote:
> >
> > Hi
> >
> > it looks like one recent commit causes the kernel build to break.
> >
> > GEN scripts/gdb/linux/constants.py
> > In file included from /usr/include/stdlib.h:1013,
> > from arch/um/os-Linux/helper.c:6:
> > /usr/include/x86_64-linux-gnu/bits/stdlib-float.h: In function ‘atof’:
> > /usr/include/x86_64-linux-gnu/bits/stdlib-float.h:26:1: error: SSE register
> > return with SSE disabled
> > 26 | {
> > | ^
> > CC init/noinitramfs.o
> >
> > It works again by reverting commit 884981867947 ("rust: arch/um:
> > Disable FP/SIMD instruction to match x86").
> >
> > $ cat /proc/cpuinfo
> >
> > [...]
> >
> > processor : 31
> > vendor_id : AuthenticAMD
> > cpu family : 23
> > model : 49
> > model name : AMD Ryzen Threadripper PRO 3955WX 16-Cores
> >
> > Thanks
> >
> > Roberto
> >
>
> Yup, this is definitely caused by 884981867947 ("rust: arch/um:
> Disable FP/SIMD instruction to match x86").
>
> It looks like GCC fixed this upstream in GCC 11:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99652
>
> Equally though, not having this change breaks older versions of LLVM,
> so I think it's best to not totally revert it.
>
> Something like this should work:
> ---
> diff --git a/arch/x86/Makefile.um b/arch/x86/Makefile.um
> index d2e95d1d4db7..1aa64846e539 100644
> --- a/arch/x86/Makefile.um
> +++ b/arch/x86/Makefile.um
> @@ -3,9 +3,14 @@ core-y += arch/x86/crypto/
>
> #
> # Disable SSE and other FP/SIMD instructions to match normal x86
> +# This is required to work around issues in older LLVM versions, but breaks
> +# GCC versions < 11. See:
> +# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99652
> #
> +ifeq ($(CONFIG_CC_IS_CLANG),y)
> KBUILD_CFLAGS += -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx
> KBUILD_RUSTFLAGS +=
> -Ctarget-feature=-sse,-sse2,-sse3,-ssse3,-sse4.1,-sse4.2,-avx,-avx2
> +endif
>
> ifeq ($(CONFIG_X86_32),y)
> START := 0x8048000
> ---
I also encountered the same issue, and confirmed this fixes the issue on my
machine.
Tested-by: SeongJae Park <s...@kernel.org>
Thanks,
SJ
>
> Though the more correct solution would be to just check
> gcc-min-version || clang. That probably requires a bit more
> makefile-fu to do cleanly, though.
>
> Cheers,
> -- David
_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um
