Re: [PATCH 00/14] Modularize schedutil

[Rafael J. Wysocki]

On Tue, May 12, 2020 at 6:26 PM Quentin Perret <qperret@xxxxxxxxxx> wrote:
>
> On Tuesday 12 May 2020 at 17:30:36 (+0200), Rafael J. Wysocki wrote:
> > On Tue, May 12, 2020 at 5:11 PM Quentin Perret <qperret@xxxxxxxxxx> wrote:
> > > The end goal with GKI is the following: Google will release a single
> > > binary kernel image (signed, etc etc) that all devices using a given
> > > Android version will be required to use. That image is however going to
> > > be only for the core of the kernel (no drivers or anything of the sort).
> > > Vendors and OEMs, on their end, will be responsible to build and ship
> > > GKI-compatible modules for their respective devices. So, Android devices
> > > will eventually ship with a Google-issued GKI, plus a bunch of
> > > vendor-provided modules loaded during boot.
> >
> > If that is the case, then I absolutely think that schedutil should be
> > part of the GKI.
> >
> > Moreover, that would have been my opinion even if it had been modular
> > in the first place.
>
> I definitely understand the feeling. Heck I contributed to schedutil, so
> I'd love to see the entire world run it :-)
>
> But my personal preference doesn't seem to matter in this world, sadly.
> The truth is, we cannot afford to be arbitrary in our decisions in GKI.
> Switching governors and such is a fully supported feature upstream, and
> it has been for a long time. Taking that away from partners is not the
> goal, nor the intention, of GKI.

It still will be possible with schedutil built-in, however.

> They will be able to choose whatever
> governor they want, because there are no *objective* reasons to not let
> them do that.

Which, again, is still possible with non-modular schedutil AFAICS.

I don't see any technical reason for making schedutil modular in the
context of GKI other than to make the GKI image smaller, but I don't
expect that to be significant enough.

Is there anything else I am missing?

> > > This is a significant shift from the current model where vendors
> > > completely own the kernel, and are largely free to use the kernel config
> > > they want. Today, those who don't use schedutil are free to turn the
> > > config off, for example.
> >
> > So why is this regarded as a good thing?
>
> You mean using something else than schedutil?

I mean why allowing people to compile schedutil out is regarded as a good thing.

> It is not seen as a good
> thing at all, at least not by me. But we have the same problem as
> upstream. We cannot remove the other governors or the governor API for a
> simple reason: they have users :/

I'm not saying about removing any of that.  I'm just trying to
understand why you need schedutil to be modular so as to make those
things possible.

Cheers!