From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932779AbdJaLv6 (ORCPT ); Tue, 31 Oct 2017 07:51:58 -0400 Received: from usa-sjc-mx-foss1.foss.arm.com ([217.140.101.70]:34680 "EHLO foss.arm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932413AbdJaLvz (ORCPT ); Tue, 31 Oct 2017 07:51:55 -0400 Date: Tue, 31 Oct 2017 11:51:56 +0000 From: Will Deacon To: "Jason A. Donenfeld" Cc: linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, mark.rutland@arm.com Subject: Re: [PATCH] arm64: support __int128 on gcc 5+ Message-ID: <20171031115156.GF5584@arm.com> References: <20171031114319.24436-1-Jason@zx2c4.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20171031114319.24436-1-Jason@zx2c4.com> User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Oct 31, 2017 at 12:43:19PM +0100, Jason A. Donenfeld wrote: > Versions of gcc prior to gcc 5 emitted a __multi3 function call when > dealing with TI types, resulting in failures when trying to link to > libgcc, and more generally, horrible performance. However, since gcc 5, > the compiler supports actually emitting fast instructions, which means > we can at long last enable this option and receive the speedups. > > The gcc commit that added proper Aarch64 support is: > https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=d1ae7bb994f49316f6f63e6173f2931e837a351d > > This commit appears to be part of the gcc 5 release. > > Signed-off-by: Jason A. Donenfeld > --- > arch/arm64/Makefile | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/arch/arm64/Makefile b/arch/arm64/Makefile > index 939b310913cf..1f8a0fec6998 100644 > --- a/arch/arm64/Makefile > +++ b/arch/arm64/Makefile > @@ -53,6 +53,8 @@ KBUILD_AFLAGS += $(lseinstr) $(brokengasinst) > KBUILD_CFLAGS += $(call cc-option,-mabi=lp64) > KBUILD_AFLAGS += $(call cc-option,-mabi=lp64) > > +KBUILD_CFLAGS += $(call cc-ifversion, -ge, 0500, -DCONFIG_ARCH_SUPPORTS_INT128) > + > ifeq ($(CONFIG_CPU_BIG_ENDIAN), y) > KBUILD_CPPFLAGS += -mbig-endian > CHECKFLAGS += -D__AARCH64EB__ Which code in the kernel actually uses 128-bit types directly? I know we have some unfortunate occurences in our headers (including uapi) for the vector registers, but I thought we generally used asm or copy routines to access those. Are you seeing a performance issue without this? Will