Adding tree to kernelci

Adding tree to kernelci

[Linus Walleij]

Hi,

could the GPIO tree:
https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/

Branches "devel", "fixes" and "for-next" (no others!) be added to
the kernelci loop?

Yours,
Linus Walleij

Re: Adding tree to kernelci

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 489 bytes --]

On Thu, Jan 31, 2019 at 12:50 PM Linus Walleij <linus.walleij@linaro.org>
wrote:

> Hi,
>
> could the GPIO tree:
> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
>
> Branches "devel", "fixes" and "for-next" (no others!) be added to
> the kernelci loop?
>

We have the capacity to build them so I'm in favour of it.

Out of interest, are there any existing tests that can be run to
cover the GPIO subsystem and which we could consider adding to
KernelCI?

Guillaume

[-- Attachment #2: Type: text/html, Size: 1102 bytes --]

Re: Adding tree to kernelci

[Matt Hart]

On Thu, 31 Jan 2019 at 18:28, Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:
>
>
> On Thu, Jan 31, 2019 at 12:50 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>>
>> Hi,
>>
>> could the GPIO tree:
>> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
>>
>> Branches "devel", "fixes" and "for-next" (no others!) be added to
>> the kernelci loop?

How often are they updated?

>
>
> We have the capacity to build them so I'm in favour of it.
>
> Out of interest, are there any existing tests that can be run to
> cover the GPIO subsystem and which we could consider adding to
> KernelCI?
>
> Guillaume
> 

Re: Adding tree to kernelci

[Linus Walleij]

On Thu, Jan 31, 2019 at 7:28 PM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:
> On Thu, Jan 31, 2019 at 12:50 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>>
>> could the GPIO tree:
>> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
>>
>> Branches "devel", "fixes" and "for-next" (no others!) be added to
>> the kernelci loop?
>
>
> We have the capacity to build them so I'm in favour of it.
>
> Out of interest, are there any existing tests that can be run to
> cover the GPIO subsystem and which we could consider adding to
> KernelCI?

The blackbox (subsystem core) tests we have are done from
userspace using the tests in Bartosz' libgpiod:
https://git.kernel.org/pub/scm/libs/libgpiod/libgpiod.git/tree/tests

These will (AFAIK) probe the gpio-mockup.c device which
reflects its internal state out in debugfs, and thus it closes
the loop on the core and test it.

There has been some flux in the kernel parts lately as it
is being refactored/improved.

I think Anders is looking into enabling these tests.

Yours,
Linus Walleij

Re: Adding tree to kernelci

[Linus Walleij]

On Thu, Jan 31, 2019 at 7:37 PM Matt Hart <matthew.hart@linaro.org> wrote:
> On Thu, 31 Jan 2019 at 18:28, Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:

> >> could the GPIO tree:
> >> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
> >>
> >> Branches "devel", "fixes" and "for-next" (no others!) be added to
> >> the kernelci loop?
>
> How often are they updated?

Depends on my work schedule, I suppose weekly to biweekly
but when I focus on it they change several times the same
day.

The best would be that the tree be checked once per 24 hrs
at my nighttime for if any changes done during the day.
I think that would seldom amount to it running two days in
a row except during intense times (merge window).

Yours,
Linus Walleij

Re: Adding tree to kernelci

[Bartosz Golaszewski]

pt., 1 lut 2019 o 13:15 Linus Walleij <linus.walleij@linaro.org> napisał(a):
>
> On Thu, Jan 31, 2019 at 7:28 PM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
> > On Thu, Jan 31, 2019 at 12:50 PM Linus Walleij <linus.walleij@linaro.org> wrote:
> >>
> >> could the GPIO tree:
> >> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
> >>
> >> Branches "devel", "fixes" and "for-next" (no others!) be added to
> >> the kernelci loop?
> >
> >
> > We have the capacity to build them so I'm in favour of it.
> >
> > Out of interest, are there any existing tests that can be run to
> > cover the GPIO subsystem and which we could consider adding to
> > KernelCI?
>
> The blackbox (subsystem core) tests we have are done from
> userspace using the tests in Bartosz' libgpiod:
> https://git.kernel.org/pub/scm/libs/libgpiod/libgpiod.git/tree/tests
>
> These will (AFAIK) probe the gpio-mockup.c device which
> reflects its internal state out in debugfs, and thus it closes
> the loop on the core and test it.
>
> There has been some flux in the kernel parts lately as it
> is being refactored/improved.
>
> I think Anders is looking into enabling these tests.
>

Hello,

yes we have a set of userspace tests that can be run if you can enable
the gpio-mockup testing module. I can help with building the suite -
it requires udev and kmod to work.

Also: Linus is right - we've had a fix applied to the gpiolib tree
last release that broke a single test-case and we're now in progress
of reworking the entire testing interface. I hope to get it in for
v5.1 and then I'd need to rework the user-space part as well.

Let me know if you need anything else.

Best regards,
Bartosz Golaszewski

Re: Adding tree to kernelci

[Anders Roxell]

On Fri, 1 Feb 2019 at 13:15, Linus Walleij <linus.walleij@linaro.org> wrote:
>
> On Thu, Jan 31, 2019 at 7:28 PM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
> > On Thu, Jan 31, 2019 at 12:50 PM Linus Walleij <linus.walleij@linaro.org> wrote:
> >>
> >> could the GPIO tree:
> >> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
> >>
> >> Branches "devel", "fixes" and "for-next" (no others!) be added to
> >> the kernelci loop?
> >
> >
> > We have the capacity to build them so I'm in favour of it.
> >
> > Out of interest, are there any existing tests that can be run to
> > cover the GPIO subsystem and which we could consider adding to
> > KernelCI?
>
> The blackbox (subsystem core) tests we have are done from
> userspace using the tests in Bartosz' libgpiod:
> https://git.kernel.org/pub/scm/libs/libgpiod/libgpiod.git/tree/tests
>
> These will (AFAIK) probe the gpio-mockup.c device which
> reflects its internal state out in debugfs, and thus it closes
> the loop on the core and test it.
>
> There has been some flux in the kernel parts lately as it
> is being refactored/improved.
>
> I think Anders is looking into enabling these tests.

Yes, that's on my list.

Cheers,
Anders

Re: Adding tree to kernelci

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 1176 bytes --]

On Fri, Feb 1, 2019 at 12:18 PM Linus Walleij <linus.walleij@linaro.org>
wrote:

> On Thu, Jan 31, 2019 at 7:37 PM Matt Hart <matthew.hart@linaro.org> wrote:
> > On Thu, 31 Jan 2019 at 18:28, Guillaume Tucker
> > <guillaume.tucker@gmail.com> wrote:
>
> > >> could the GPIO tree:
> > >>
> https://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio.git/
> > >>
> > >> Branches "devel", "fixes" and "for-next" (no others!) be added to
> > >> the kernelci loop?
> >
> > How often are they updated?
>
> Depends on my work schedule, I suppose weekly to biweekly
> but when I focus on it they change several times the same
> day.
>
> The best would be that the tree be checked once per 24 hrs
> at my nighttime for if any changes done during the day.
> I think that would seldom amount to it running two days in
> a row except during intense times (merge window).
>

I see there's now a PR to add these branches:

  https://github.com/kernelci/kernelci-core-staging/pull/100

I've added a comment about the number of builds, as at the moment
we can't easily configure to only check these branches every 24h
when all the other branches are being checked every hour.

Guillaume

[-- Attachment #2: Type: text/html, Size: 2116 bytes --]

Re: Adding tree to kernelci

[Linus Walleij]

On Mon, Feb 25, 2019 at 2:04 PM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:
> On Fri, Feb 1, 2019 at 12:18 PM Linus Walleij <linus.walleij@linaro.org> wrote:

>> The best would be that the tree be checked once per 24 hrs
>> at my nighttime for if any changes done during the day.
>> I think that would seldom amount to it running two days in
>> a row except during intense times (merge window).
>
> I see there's now a PR to add these branches:
>
>   https://github.com/kernelci/kernelci-core-staging/pull/100

Sweet!

> I've added a comment about the number of builds, as at the moment
> we can't easily configure to only check these branches every 24h
> when all the other branches are being checked every hour.

Fair enough, I don't think it will stress the system too much.

Thanks a lot!
Linus Walleij

Re: Adding tree to kernelci

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 1446 bytes --]

Hi Linus,

On Tue, Feb 26, 2019 at 3:15 PM Linus Walleij <linus.walleij@linaro.org>
wrote:

> On Mon, Feb 25, 2019 at 2:04 PM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
> > On Fri, Feb 1, 2019 at 12:18 PM Linus Walleij <linus.walleij@linaro.org>
> wrote:
>
> >> The best would be that the tree be checked once per 24 hrs
> >> at my nighttime for if any changes done during the day.
> >> I think that would seldom amount to it running two days in
> >> a row except during intense times (merge window).
> >
> > I see there's now a PR to add these branches:
> >
> >   https://github.com/kernelci/kernelci-core-staging/pull/100
>
> Sweet!
>
> > I've added a comment about the number of builds, as at the moment
> > we can't easily configure to only check these branches every 24h
> > when all the other branches are being checked every hour.
>
> Fair enough, I don't think it will stress the system too much.
>

While there is still some headroom in the build capacity, it
seems worthwhile reviewing what each maintainer tree actually
needs to build.  Some maintainers need to have full coverage, but
often it's not necessary to build more than the defconfigs for
all the architectures.

By doing this, we can enable many more maintainer trees and also
reduce the turnaround time.

Could you please confirm which combinations of architectures,
defconfigs and fragments you need for each branch in your
linux-gpio tree?

Thanks,
Guillaume

[-- Attachment #2: Type: text/html, Size: 2219 bytes --]

Re: Adding tree to kernelci

[Linus Walleij]

On Wed, Jun 5, 2019 at 3:01 PM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:

> While there is still some headroom in the build capacity, it
> seems worthwhile reviewing what each maintainer tree actually
> needs to build.  Some maintainers need to have full coverage, but
> often it's not necessary to build more than the defconfigs for
> all the architectures.
>
> By doing this, we can enable many more maintainer trees and also
> reduce the turnaround time.
>
> Could you please confirm which combinations of architectures,
> defconfigs and fragments you need for each branch in your
> linux-gpio tree?

I just need the defconfigs for sure, all architectures have GPIO
except UML and S390 so these two can be skipped. Also any
virtualization targets if there are such.

Yours,
Linus Walleij

Re: Adding tree to kernelci

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 1481 bytes --]

On Fri, Jun 7, 2019 at 9:48 PM Linus Walleij <linus.walleij@linaro.org>
wrote:

> On Wed, Jun 5, 2019 at 3:01 PM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
>
> > While there is still some headroom in the build capacity, it
> > seems worthwhile reviewing what each maintainer tree actually
> > needs to build.  Some maintainers need to have full coverage, but
> > often it's not necessary to build more than the defconfigs for
> > all the architectures.
> >
> > By doing this, we can enable many more maintainer trees and also
> > reduce the turnaround time.
> >
> > Could you please confirm which combinations of architectures,
> > defconfigs and fragments you need for each branch in your
> > linux-gpio tree?
>
> I just need the defconfigs for sure, all architectures have GPIO
> except UML and S390 so these two can be skipped. Also any
> virtualization targets if there are such.
>

Sounds good, I've made a PR to only build the defconfig of each
arch, which amounts to 6 builds:

  https://github.com/kernelci/kernelci-core/pull/118

Here are some results from a staging run with the devel branch:

  https://staging.kernelci.org/job/linusw/branch/devel/

Can you please confirm these are the builds you need, for your 3
branches?

Also I've noticed there aren't any email reports associated with
your tree, so they just get sent to a default alias instead.
Maybe at least the linux-gpio list should be added?  Let me know
and I can add that to the PR.

Guillaume

[-- Attachment #2: Type: text/html, Size: 2224 bytes --]

Re: Adding tree to kernelci

[Linus Walleij]

On Tue, Jul 9, 2019 at 12:52 PM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:

> Sounds good, I've made a PR to only build the defconfig of each
> arch, which amounts to 6 builds:

Sounds good.

> Here are some results from a staging run with the devel branch:
>
>   https://staging.kernelci.org/job/linusw/branch/devel/

This isn't working right now the page is full of error messages.

> Also I've noticed there aren't any email reports associated with
> your tree, so they just get sent to a default alias instead.
> Maybe at least the linux-gpio list should be added?  Let me know
> and I can add that to the PR.

Yes linux-gpio@vger.kernel.org should probably be added.

Yours,
Linus Walleij

Re: Adding tree to kernelci

[Guillaume Tucker]

[-- Attachment #1: Type: text/plain, Size: 1153 bytes --]

On Tue, Jul 9, 2019 at 12:27 PM Linus Walleij <linus.walleij@linaro.org>
wrote:

> On Tue, Jul 9, 2019 at 12:52 PM Guillaume Tucker
> <guillaume.tucker@gmail.com> wrote:
>
> > Sounds good, I've made a PR to only build the defconfig of each
> > arch, which amounts to 6 builds:
>
> Sounds good.


Great.

> Here are some results from a staging run with the devel branch:
> >
> >   https://staging.kernelci.org/job/linusw/branch/devel/
>
> This isn't working right now the page is full of error messages.
>

Oops, there was a broken change being tested there.  It should be
back on track now.


> > Also I've noticed there aren't any email reports associated with
> > your tree, so they just get sent to a default alias instead.
> > Maybe at least the linux-gpio list should be added?  Let me know
> > and I can add that to the PR.
>
> Yes linux-gpio@vger.kernel.org should probably be added.
>

OK I've added a path to the PR to add that, with just build and
boot reports for now.  I guess it would be useful to run some
GPIO tests at some point (libgpiod was mentioned earlier on this
thread), and then enable email reports for that.

Thanks,
Guillaume

[-- Attachment #2: Type: text/html, Size: 2177 bytes --]

Re: Adding tree to kernelci

[Linus Walleij]

On Tue, Jul 9, 2019 at 2:44 PM Guillaume Tucker
<guillaume.tucker@gmail.com> wrote:
> On Tue, Jul 9, 2019 at 12:27 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>> On Tue, Jul 9, 2019 at 12:52 PM Guillaume Tucker
>> <guillaume.tucker@gmail.com> wrote:

>> >   https://staging.kernelci.org/job/linusw/branch/devel/
>>
>> This isn't working right now the page is full of error messages.
>
> Oops, there was a broken change being tested there.  It should be
> back on track now.

Looks good!

>> Yes linux-gpio@vger.kernel.org should probably be added.
>
> OK I've added a path to the PR to add that, with just build and
> boot reports for now.  I guess it would be useful to run some
> GPIO tests at some point (libgpiod was mentioned earlier on this
> thread), and then enable email reports for that.

Yes Bartosz knows what is the proper approach here.
A set of whitebox tests from userspace is coming into place.

Yours,
Linus Walleij