From mboxrd@z Thu Jan 1 00:00:00 1970 From: Theodoros Foradis Subject: Re: [PATCH v2 0/4] Add GCC cross compiler for arm-none-eabi. Date: Thu, 29 Sep 2016 20:39:00 +0300 Message-ID: <87wphubo7f.fsf@openmailbox.org> References: <20160918063513.11027-1-rekado@elephly.net> <20160919202340.4122-1-theodoros.for@openmailbox.org> <87d1jp90l6.fsf@gnu.org> <878tubt5xd.fsf@mdc-berlin.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:45820) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bpfK1-00040K-7B for guix-devel@gnu.org; Thu, 29 Sep 2016 13:40:38 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bpfJw-0008Q2-Vg for guix-devel@gnu.org; Thu, 29 Sep 2016 13:40:36 -0400 Received: from mail.openmailbox.org ([62.4.1.34]:54764) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bpfJw-0008PX-ME for guix-devel@gnu.org; Thu, 29 Sep 2016 13:40:32 -0400 In-reply-to: <878tubt5xd.fsf@mdc-berlin.de> List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org Sender: "Guix-devel" To: guix-devel Ricardo Wurmus writes: > Ludovic Court=C3=A8s writes: > >> Hi Theodoros, >> >> Theodoros Foradis skribis: >> >>> The original patch series was working correctly and producing working= binaries. Some >>> flags (that I had been using with 6.2.0) are missing from that versio= n of GCC 4.9, >>> so I added 6.2.0 as an extra option. I have tested it to produce work= ing binaries. >>> >>> Here are some modifications to Ricardo's patches for the arm-none-eab= i >>> bare metal cross compiler. The following changes have been made: >>> >>> - I have modified xbinutils to use binutils 2.25.1 from cross-base, a= s it compiles >>> correctly with it. The version from the svn commit that was used by R= icardo is compiling >>> correct binaries as well. Thus, if it is deemed appropriate, the sour= ce for xbinutils can >>> be swapped for the previous one, with (seemingly) no difference. >>> >>> - The xgcc of the original, was failing to find the headers that newl= ib provided. >>> I have set the native-cross-paths as a workaround. Not sure if there = is a better >>> alternative, or if the failure was my mistake. >>> >>> - A package for cross GCC 6.2.0 is added, with appropriate patches fo= r multilib >>> support. >>> >>> - Newlib-arm-none-eabi and newlib-nano-arm-none-eabi have been change= d to >>> procedures, taking an xgcc as argument, so as to facilitate building = with >>> either version of gcc. >>> >>> - An arm-none-eabi-toolchain procedure is declared, to create toolcha= in packages >>> for both gcc and newlib version. The four toolchain variables follow.= Not sure >>> if it's a mistake to include "nano" in the toolchain version. >> >> This all sounds reasonable to me. Ricardo was interested in using thi= s >> toolchain for one specific purpose, so maybe we=E2=80=99ll want to che= ck that it >> also works here. Ricardo: could you comment? > > The changes seem reasonable. I wasn=E2=80=99t happy with using fixed S= VN > revisions in my patches, so I=E2=80=99m glad that this can be avoided. > > I haven=E2=80=99t yet found the time to apply the proposed changes, bui= ld the > toolchain and try it with the Axoloti board. I hope I=E2=80=99ll be ab= le try on > Sunday to first address your comments, Ludo. Then I=E2=80=99ll check t= he > suggested changes made by Theodoros (e.g. using different binutils and > doing without SVN). > > Theodoros, I see that your patch set includes some of my patches as > well. The only changes I can see is the addition of the native search > paths and parameterising newlib with xgcc, both of which I=E2=80=99ll a= dd. > After applying my modified patches I would apply your patches that add > =E2=80=9Carm-none-eabi-gcc-6=E2=80=9D and the =E2=80=9Carm-none-eabi-to= olchain=E2=80=9D. > > Is this acceptable? > > ~~ Ricardo Yes, actually I did apply my patches on top of yours. It is perfectly acceptable, with either version of binutils. Regards, --=20 Theodoros Foradis