On Fri, Oct 5, 2018 at 1:01 AM H.J. Lu <hongjiu...@intel.com> wrote:
>
>
> With
>
> gcc -O2 -fPIC -flto -g -c -o a.o a.c
> gcc -O2 -fPIC -flto -g -mavx -c -o b.o b.c
> gcc -shared -O2 -fPIC -flto -g -o lib1.so a.o b.o
>
> LTO correctly generates AVX for b.o and SSE for a.o. But the GCC driver
> passes -msse2avx to assembler, which encodes SSE instructions as AVX
> instructions. We shouldn't pass -msse2avx to assembler for -mavx.
>
> Tested on x86-64. OK for trunk?
>
> Thanks.
>
> H.J.
> ---
> PR target/87522
> * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
> assembler for -mavx.
> * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
OK for mainline and backports.
Thanks,
Uros.
> gcc/config/i386/gnu-user.h | 2 +-
> gcc/config/i386/gnu-user64.h | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/gcc/config/i386/gnu-user.h b/gcc/config/i386/gnu-user.h
> index a922c9b93fa..baed87aa54f 100644
> --- a/gcc/config/i386/gnu-user.h
> +++ b/gcc/config/i386/gnu-user.h
> @@ -67,7 +67,7 @@ along with GCC; see the file COPYING3. If not see
>
> #undef ASM_SPEC
> #define ASM_SPEC \
> - "--32 %{!mno-sse2avx:%{mavx:-msse2avx}} %{msse2avx:%{!mavx:-msse2avx}}"
> + "--32 %{msse2avx:%{!mavx:-msse2avx}}"
>
> #undef SUBTARGET_EXTRA_SPECS
> #define SUBTARGET_EXTRA_SPECS \
> diff --git a/gcc/config/i386/gnu-user64.h b/gcc/config/i386/gnu-user64.h
> index f7a68fdecf0..09141ce3508 100644
> --- a/gcc/config/i386/gnu-user64.h
> +++ b/gcc/config/i386/gnu-user64.h
> @@ -50,7 +50,7 @@ see the files COPYING3 and COPYING.RUNTIME respectively.
> If not, see
> #define ASM_SPEC "%{" SPEC_32 ":--32} \
> %{" SPEC_64 ":--64} \
> %{" SPEC_X32 ":--x32} \
> - %{!mno-sse2avx:%{mavx:-msse2avx}} %{msse2avx:%{!mavx:-msse2avx}}"
> + %{msse2avx:%{!mavx:-msse2avx}}"
>
> #define GNU_USER_TARGET_LINK_SPEC \
> "%{" SPEC_64 ":-m " GNU_USER_LINK_EMULATION64 "} \
> --
> 2.17.1
>
