From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:42631) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1f9RQa-000120-CA for qemu-devel@nongnu.org; Fri, 20 Apr 2018 04:29:57 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1f9RQX-0007e5-8O for qemu-devel@nongnu.org; Fri, 20 Apr 2018 04:29:56 -0400 Received: from mail-wr0-x244.google.com ([2a00:1450:400c:c0c::244]:36176) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1f9RQX-0007dU-0l for qemu-devel@nongnu.org; Fri, 20 Apr 2018 04:29:53 -0400 Received: by mail-wr0-x244.google.com with SMTP id m26-v6so2822473wrb.3 for ; Fri, 20 Apr 2018 01:29:52 -0700 (PDT) References: <20180419135901.30035-1-alex.bennee@linaro.org> <20180419135901.30035-32-alex.bennee@linaro.org> <8ccf1143-5895-ca8b-eea5-c7d96c992455@amsat.org> <87vacn9jd5.fsf@linaro.org> <60b519a7-f49b-95e7-b9db-3e74133cf8c0@amsat.org> From: Alex =?utf-8?Q?Benn=C3=A9e?= In-reply-to: <60b519a7-f49b-95e7-b9db-3e74133cf8c0@amsat.org> Date: Fri, 20 Apr 2018 09:29:50 +0100 Message-ID: <87o9ie9t6p.fsf@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v2 31/43] tests/tcg: enable building for MIPS List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Philippe =?utf-8?Q?Mathieu-Daud=C3=A9?= Cc: Peter Maydell , "Emilio G. Cota" , Fam Zheng , "Daniel P. Berrange" , Richard Henderson , andrzej zaborowski , Aurelien Jarno , Alexander Graf , QEMU Developers , Yongbok Kim Philippe Mathieu-Daud=C3=A9 writes: > On 04/19/2018 02:58 PM, Peter Maydell wrote: >> On 19 April 2018 at 18:49, Alex Benn=C3=A9e wro= te: >>> Philippe Mathieu-Daud=C3=A9 writes: >>>> On 04/19/2018 10:58 AM, Alex Benn=C3=A9e wrote: >>>>> This doesn't add any additional tests but enables building the >>>>> multiarch tests for MIPS using docker cross compilers. We don't have a >>>>> cross compiler for mips64 big endian though. >>>> >>>> Oh we have one, using CFLAGS+=3D-EB >>>> >>>> we don't have cross libraries although. >>> >>> Yeah I thought the same with the ARM compilers (-mbig-endian) but it's >>> the libraries that let us down. If, as you say, newlib gets this right I >>> suspect we should use that to build the compilers not supported out of >>> the box by Debian. >> >> I don't think we really want to get into the business of >> building our own cross compilers if we can avoid it...it's harder >> than it looks and we would be essentially reinventing the wheel. > > I totally agree. > > I'v been looking at existing maintained images we can use to > cross-build, so we only have to pull (download) and image and use it, > not wasting cpu time building it. > > However some images might be huge, full of things we don't need. But I > prefer that rather than having to maintain cross toolchains. > > An example I like to use is the coreboot-sdk, designed to build > i386/amd64 binaries, it fulfills our needs. But takes 2GB... > > $ docker pull coreboot/coreboot-sdk:1.47 > 1.47: Pulling from coreboot/coreboot-sdk > 3d6aedfc3e47: Downloading [> ] 538.2kB/1.201GB > d23c9a72f1e5: Waiting > > Another candidate is crossbuild, ~1GB > $ docker pull multiarch/crossbuild I agree to a point. Our current images are all based off "official" docker hub images. I'm probably giving the state of officialness too much credit but I'm at least reasonably certain the Debian images can be re-created ourselves if we wanted to and have publicly available recipes. As a bonus, being Debian based, we meet our GPL compliance because everything on the image can have its source pulled with apt. I'm unsure about how the coreboot image was built or indeed if it is an official image. From my point of view it is just a bunch of binaries in a set of docker layers. So personally I would put "build our own compiler from source inside docker" slightly above "unpack a random binary tarball into docker". That said with some of our funkier supported guests we may end up having to do just that :-/ -- Alex Benn=C3=A9e