Re: RISC-V

Hi Alexandre,

On Tue, Oct 1, 2024 at 12:31 PM Gustavo Padovan <[email protected]> wrote:
>
>
> Hi Alexandre,
>
> ---- On Tue, 01 Oct 2024 12:21:49 -0300 Alexandre Ghiti  wrote ---
>
>  > Hi everyone,
>  >
>  > We (the RISC-V community) are looking for improving our CIs that are
>  > currently spread over all the RISC-V companies, mostly private and very
>  > incomplete.
>
> That is great to hear!  We have other hardware vendors coming to KernelCI with the same wishes of improving CI overall.
>
>  >
>  > I only found one RISC-V build in kernelci, can I do something to help
>  > adding a better RISC-V support?
>
> YES!
>
> First of all, we can enable more kernel trees and kernel configs in the maestro system.
> That would give you wider build coverage. If you have a list of configs we can work on adding these.
> - https://docs.kernelci.org/kernel-community/
>
> Then, we have to look at running these kernel on actual RISC-V devices. We can get RISC-V devices in KernelCI through a maestro lab, either through some of the existing labs or setting up a dedicated lab. Cc'ing Rafael from RISC-V as we discussed that already.
> - https://docs.kernelci.org/labs/
>
> Another way, that could be relatively easy if you have a CI setup already, is to get the results of the current CIs in the RISC-V ecosystem into KCIDB (as long as you area running fully Open Source software in those tests).
> - https://docs.kernelci.org/kcidb/
>
> Let us know how does that sound to you. We would happy to learn more about the ways we can make this work and chat more with you.

In addition to everything Gustavo mentioned above, Red Hat's Fedora
project can be another path as it uses CKI to do some CI testing and
submit results to the same KCIDB mentioned above.  I believe Red Hat
already has folks in the RISC-V community working on getting the
hardware enabled on Fedora and possibly RHEL.  Just another option.
Either way works or both!

Cheers,
Don

>
> Best,
>
> - Gus
>
>