Re: Kselftest implementation


Alice Ferrazzi
 

Hi,

thanks for the review,


On Thu, Apr 22, 2021 at 4:35 PM Nobuhiro Iwamatsu <nobuhiro1.iwamatsu@...> wrote:
Hi,

Thanks for your work.

On Thu, Apr 15, 2021 at 11:04:26AM +0900, Alice Ferrazzi wrote:
> Hello Chris,
>
> On Wed, Apr 14, 2021 at 7:27 AM Chris Paterson <Chris.Paterson2@...>
> wrote:
>
> > Hello Alice,
> >
> > > From: cip-dev@... <cip-dev@...> On
> > > Behalf Of Alice Ferrazzi via lists.cip-project.org
> > > Sent: 13 April 2021 18:30
> > >
> > > Hi everyone,
> > >
> > > I'm Alice Ferrazzi from Cybertrust Japan.
> > >
> > > I'm currently working on adding Kselftest to the current CIP CI tests of
> > linux-
> > > cip Kernel
> >
> > Thank you for your work on this!
> >
> > >
> > > I made and tested kselftest on a testing instance of lava, currently
> > tested
> > > only on qemu x86_64
> >
> > If you'd like to test on a wider range of platforms then get in touch and
> > we'll get you access to the CIP LAVA instance.
> >
>
> This would be nice!
>
>
> > > I implemented a simple skiplist in the kselftest envelope for skipping
> > tests
> > > that are currently not working.
> > > The usage is similar to how spectre-meltdown-checker-test is currently
> > > working.
> > >
> > > Raw log:
> > >
> > > http://140.211.166.173:10080/scheduler/job/329
> > > Results list:
> > > http://140.211.166.173:10080/results/329/1_kselftest
> >
> > It looks like a great start.
> >
> >
> Thanks
>
> >
> > >
> > > My idea is to start implementing kselftest tests for at least qemu x86_64
> > > architecture and expand in the future to other architectures.
> > > As Iwamatsu-san pointed, we need also to create kselftest binary.
> > >
> > > I'm currently thinking of implementing kselftest binary creation
> > mechanism
> > > on linux-cip-ci kernel build script but I still didn't start to work on
> > this yet.
> > >
> > > Kselftest test code:
> > >
> > > https://gitlab.com/alice.ferrazzi/cip-kselftest
> >
> >
> Is there any work we can leverage from the Linaro test-definitions repo?
> >
> > https://github.com/Linaro/test-definitions/tree/master/automated/linux/kselftest
> >
> > That was the first thing I tried out but was not working correctly on the
> cip rootfs enviroment

As you write, cip rootfs lacks the software to run kselftest. Therefore,
it is necessary to make a separate flavor of rootfs.
There are also two types of cip rootfs, and we need to use either or
both for kernel testing.
 

ok.

 
>
>
> > KernelCI also recently implemented kselftest support if that's useful at
> > all, this'll include the build side of things.
> > https://github.com/kernelci/kernelci-core/tree/main/config/lava/kselftest
> >
> >
> I know about KernelCI effort for kselftest and I think to use a similar
> approach about the kselftest binary creation implementation.
>
> The code I wrote is based on Glava-tests that is part of GKernelCI.
> GKernelCI is the current Gentoo kernel CI system that is reporting and
> collaborating with KernelCI.
> https://github.com/GKernelCI/Glava-tests
> Using a different repository currently is helpful for self managing the
> skiplist, but it could be changed in the future.


I think this needs to be done first.
I also think there are several approaches to creating kselftest
binaries.

- Continue to use a particular kernel version of ksefltests.
- Use CIP kernel kselftest.
- Use the latest kernel kselftest.

I used to build an environment using the latter.

https://gitlab.com/cip-project/cip-kernel/linux-cip/-/tree/ci/iwamatsu/linux-cip-kselftest

By the way, do you know which kernelci they are using?


KernelCI is building kselftest in production for at least linux-next and mainline any time a new revision is discovered on the branches. on staging.kernelci.org kselftest is built for each job.

Currently I'm working on a similar implementation for CIP testing.

Thanks,
Alicef



--
======================================
Cybertrust Japan Co.,Ltd.
Alice Ferrazzi
alice.ferrazzi@...
======================================

Join cip-dev@lists.cip-project.org to automatically receive all group messages.