[igt-dev] Testing s2idle/S0?

[igt-dev] Testing s2idle/S0?

[Lyude Paul]

Hi! Recently one of my coworkers from QA (cc'd here) asked me about whether we
had any tests for s2idle in igt - to which I looked, didn't see anything that
looked like a match, and suggested that they consider contributing some tests
for it. Note that when I mention s2idle here, I'm thinking about systems which
support both s2idle and S3 - as I know that s2idle is already being used by
default for suspend/resume by igt on s2idle only systems.

This made me realize that we don't actually have any explicit suspend/resume
tests - we only have tests which handle testing other functionality across
suspend/resume. This seems like it makes sense for the most part, which brings
me to my real question here - would it be worth it to try adding s2idle
support for most igt tests, so that on s2idle/S3 hybrid systems both suspend
modes are tested individually for each suspend/resume test? 
-- 
Cheers,
 Lyude Paul (she/her)
 Software Engineer at Red Hat

_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev

Re: [igt-dev] Testing s2idle/S0?

[Daniel Vetter]

On Wed, Jul 28, 2021 at 10:06 PM Lyude Paul <lyude@redhat.com> wrote:
> Hi! Recently one of my coworkers from QA (cc'd here) asked me about whether we
> had any tests for s2idle in igt - to which I looked, didn't see anything that
> looked like a match, and suggested that they consider contributing some tests
> for it. Note that when I mention s2idle here, I'm thinking about systems which
> support both s2idle and S3 - as I know that s2idle is already being used by
> default for suspend/resume by igt on s2idle only systems.
>
> This made me realize that we don't actually have any explicit suspend/resume
> tests - we only have tests which handle testing other functionality across
> suspend/resume. This seems like it makes sense for the most part, which brings
> me to my real question here - would it be worth it to try adding s2idle
> support for most igt tests, so that on s2idle/S3 hybrid systems both suspend
> modes are tested individually for each suspend/resume test?

If you really add it to all of them the runtime explosion is likely
prohibitive. But a handful of explicit ones should be fine I guess.
-Daniel

> --
> Cheers,
>  Lyude Paul (she/her)
>  Software Engineer at Red Hat
>
> _______________________________________________
> igt-dev mailing list
> igt-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/igt-dev



-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev

Re: [igt-dev] Testing s2idle/S0?

[Porygon]

[-- Attachment #1: Type: text/plain, Size: 3046 bytes --]

Hey Daniel,

Sorry for the long wait for a response, QE has been rather busy :)

This will be my first time contributing directly to an upstream project, so
I may need some assistance with the whole process. I do think it would be
good to write some tests explicitly for s2idle testing since some machines
are shipped with both "deep" (this is S3, I believe) and "s2idle" supported
as suspend states. s2idle is becoming more common on laptops being shipped
so we do want to make sure this is covered. Some machines seem to support
only s2idle and are dropping support for deep; I think that the regular
suspend cases may still cover s2idle in this case specifically, so perhaps
it makes sense to have some explicit cases for the alternate state on
machines that support both, just some simple sanity checks to catch
possible regressions and ensure suspend works in general.

I know we can blacklist tests when running IGT but I'm guessing under a
typical use case, we don't want to blacklist anything by default in a
regular test run, hence the worry about extra overhead. If this is
incorrect & having some side set then we could blacklist the extra tests by
default and only have them run as a set if the user chooses to specifically.

If you have any further advice or thoughts on this, please let me know!
Just hoping to get working on this ASAP so I've got a solution in mind
within the next few weeks.


Best regards,
-- Porygon


On Thu, Jul 29, 2021 at 3:43 AM Daniel Vetter <daniel.vetter@ffwll.ch>
wrote:

> On Wed, Jul 28, 2021 at 10:06 PM Lyude Paul <lyude@redhat.com> wrote:
> > Hi! Recently one of my coworkers from QA (cc'd here) asked me about
> whether we
> > had any tests for s2idle in igt - to which I looked, didn't see anything
> that
> > looked like a match, and suggested that they consider contributing some
> tests
> > for it. Note that when I mention s2idle here, I'm thinking about systems
> which
> > support both s2idle and S3 - as I know that s2idle is already being used
> by
> > default for suspend/resume by igt on s2idle only systems.
> >
> > This made me realize that we don't actually have any explicit
> suspend/resume
> > tests - we only have tests which handle testing other functionality
> across
> > suspend/resume. This seems like it makes sense for the most part, which
> brings
> > me to my real question here - would it be worth it to try adding s2idle
> > support for most igt tests, so that on s2idle/S3 hybrid systems both
> suspend
> > modes are tested individually for each suspend/resume test?
>
> If you really add it to all of them the runtime explosion is likely
> prohibitive. But a handful of explicit ones should be fine I guess.
> -Daniel
>
> > --
> > Cheers,
> >  Lyude Paul (she/her)
> >  Software Engineer at Red Hat
> >
> > _______________________________________________
> > igt-dev mailing list
> > igt-dev@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/igt-dev
>
>
>
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch
>
>

[-- Attachment #2: Type: text/html, Size: 4013 bytes --]