[RFC 1/1] tools: Dockerfile for armv7
Tom Rini
trini at konsulko.com
Fri Nov 11 00:23:52 CET 2022
On Thu, Nov 10, 2022 at 01:40:35PM -0700, Simon Glass wrote:
> Hi,
>
> On Fri, 28 Oct 2022 at 13:19, Tom Rini <trini at konsulko.com> wrote:
> >
> > On Fri, Oct 28, 2022 at 09:14:13PM +0200, Heinrich Schuchardt wrote:
> > > On 10/28/22 21:04, Tom Rini wrote:
> > > > On Sun, Oct 16, 2022 at 12:03:30PM +0200, Heinrich Schuchardt wrote:
> > > > > On 10/16/22 09:43, Heinrich Schuchardt wrote:
> > > > > > Provide an armv7 docker environment to build and test the sandbox.
> > > > > >
> > > > > > Signed-off-by: Heinrich Schuchardt<heinrich.schuchardt at canonical.com>
> > > > > > ---
> > > > > > I have been discussing with Simon, if we can test sandbox_defconfig
> > > > > > build on arm in a docker container. This RFC is just a small step on
> > > > > > the path to it.
> > > > >
> > > > > We also need a separate requirements.txt with pygit2==1.6.1 due the
> > > > > different version of libgit in the armhf version of Ubuntu Jammy.
> > > >
> > > > So, what are the general best practices to have a Dockerfile that can be
> > > > / is used for generating multiple platforms worth of images? We should
> > > > aim for a single Dockerfile that can be used for x86-64, armv7 and
> > > > aarch64.
> > >
> > > As we execute bash commands in the Dockerfile we could use if/fi and
> > > case/esac to work around differences between architectures.
> >
> > For unavoidable differences, yes, OK.
> >
> > > Except on amd64 we only want to execute the sandbox. So there is no need to
> > > build GRUB, QEMU, and swtmp/libtmps.
> >
> > Why? There's no reason one shouldn't be able to run the whole of CI on
> > aarch64 hosts, unless there's QEMU bugs, or similar issues. The first
> > use might be just to do sandbox runs but I don't want to limit things
> > without strong reason ($X doesn't work, would be a good reason to
> > limit).
>
> I'm not sure where this is going, but how do we build sandbox on ARM?
> Do we need to use the cross compiler? I'd love to see this happen.
I'm going to keep an eye on the billing tab, but, I believe I have now
signed up for a free and reasonably spec'd arm64 oracle cloud compute
image and installed the gitlab runner on it. We'll want to put some
additional logic in the gitlab-ci.yml file so that only certain jobs
would run on runners with the lets say aarch64 tag, but, here's how we
do it I believe.
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20221110/963089b8/attachment.sig>
More information about the U-Boot
mailing list