In preparation for adding some livepatch-build-tools test update the Alpine
container to also install elfutils-dev, coreutils and GNU awk.

Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
---
I don't very much like to add coreutils and gawk, as it's also good to test
that we can build Xen with Busybox, but I also got tired of adjusting
livepatch-build-tools.
---
 automation/build/alpine/3.18.dockerfile | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/automation/build/alpine/3.18.dockerfile b/automation/build/alpine/3.18.dockerfile
index 4ae9cb5e9e30..fa6789347d87 100644
--- a/automation/build/alpine/3.18.dockerfile
+++ b/automation/build/alpine/3.18.dockerfile
@@ -47,3 +47,7 @@ RUN apk --no-cache add \
   libcap-ng-dev \
   ninja \
   pixman-dev \
+  # livepatch-tools deps
+  elfutils-dev \
+  coreutils \
+  gawk \
-- 
2.43.0

On 23/11/2023 11:23 am, Roger Pau Monne wrote:
> In preparation for adding some livepatch-build-tools test update the Alpine
> container to also install elfutils-dev, coreutils and GNU awk.
>
> Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
> ---
> I don't very much like to add coreutils and gawk, as it's also good to test
> that we can build Xen with Busybox, but I also got tired of adjusting
> livepatch-build-tools.

Right now, the alpine environment is the main one which spots violations
of Xen's requirement for simply a POSIX-compliant awk, which I think
this would break?

How much effort would it be fix livepatch-build-tools?  I think that
would be preferable, and could be persuaded to do some simple busywork...

What about coreutils?  Presumably that's down to some differences from
busybox ?

~Andrew

On Thu, Nov 23, 2023 at 11:51:33AM +0000, Andrew Cooper wrote:
> On 23/11/2023 11:23 am, Roger Pau Monne wrote:
> > In preparation for adding some livepatch-build-tools test update the Alpine
> > container to also install elfutils-dev, coreutils and GNU awk.
> >
> > Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
> > ---
> > I don't very much like to add coreutils and gawk, as it's also good to test
> > that we can build Xen with Busybox, but I also got tired of adjusting
> > livepatch-build-tools.
> 
> Right now, the alpine environment is the main one which spots violations
> of Xen's requirement for simply a POSIX-compliant awk, which I think
> this would break?

Likely, we also test on Cirrus using FreeBSD, so we would at least
spot instances where the extensions are not implemented by BSD sed.

> How much effort would it be fix livepatch-build-tools?  I think that
> would be preferable, and could be persuaded to do some simple busywork...

I can give it a try.

> What about coreutils?  Presumably that's down to some differences from
> busybox ?

Yeah, it's for the usage of `readlink -m` by livepatch-build-tools.

We will have to switch to using `readlink -f`, as that's the only
canonicalize option supported by BusyBox readlink.  Note that anyway
`readlink` is not part of POSIX.

Thanks, Roger.