Re: [PATCH v3 2/6] CI: avoid repacking initrd as part of the test job

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 14/04/2025 5:57 pm, Anthony PERARD wrote:
> On Mon, Apr 14, 2025 at 12:08:59PM +0100, Andrew Cooper wrote:
>> From: Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>
>>
>> Use the new test-artifacts which provide rootfs.cpio.gz rather than
>> initrd.tar.gz.  rootfs.cpio.gz also has all the necessary top-level
>> directories, and includes the rc_verbose setting, so these modifications can
>> be dropped.
>>
>> Having that, do not repack the whole initrd, but only pack modified
>> files and rely on Linux handling of concatenated archives.
>> This allows packing just test-related files (which includes the whole
>> toolstack), instead of the whole initrd.
>>
>> For xilinx-smoke-dom0-x86_64.sh, this involves instructing grub not to unzip
>> the archive, as doing so corrupts it.
>>
>> Signed-off-by: Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>
>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
> This patch seems to depends on the other patch series for
> "test-artifacts" repo.

Yes, there's a large interlink, although the other series is fully
committed now.

>
> I've tried to compare the rootfs fs generated by this repo, and the one
> generated in the "test-artifacts" repo, and I think there's a few
> changes, at least for the arm64 rootfs as the changes have probably
> already sailed for x86_64, namely:
>
>     Those don't exist in the new rootfs:
>         rc-update add networking sysinit
>         rc-update add modloop sysinit
>         rc-update add modules boot
>         rc-update add sysctl boot
>     There's `echo > /etc/modules` been added.
>
> But I guess none of those matter, as it already works on x86_64 tests.

Those changes are discussed in the other series, in the relevant patches.

modules is deliberately removed; we don't build them (properly), and the
scan on boot puts errors onto the console.  Emptying /etc/modules is to
remove af_packet and ipv6, again because we don't have those as modules
(we build ipv6 in).

Modloop seems to be scanning other devices for filesystems.  networking
depends on sysctl, but I can't see any of them being used, and the arm64
tests work fine without them

>
> Overall, looks good, more common code between tests!

Indeed.  We can always put them back in if they turn out to be needed.

> Reviewed-by: Anthony PERARD <anthony.perard@xxxxxxxxxx>

Thanks.

~Andrew