Re: [PATCH 03/18] soc: amlogic: Move power-domain drivers to the genpd dir

From: Bjorn Andersson
Date: Thu Jul 13 2023 - 23:07:34 EST


On Thu, Jul 13, 2023 at 11:50:29AM +0200, Ulf Hansson wrote:
> On Thu, 13 Jul 2023 at 02:45, Bjorn Andersson <andersson@xxxxxxxxxx> wrote:
> >
> > On Mon, Jul 10, 2023 at 02:45:12PM +0200, Arnd Bergmann wrote:
> > > On Mon, Jul 10, 2023, at 14:37, Ulf Hansson wrote:
> > > > On Sun, 9 Jul 2023 at 04:21, Bjorn Andersson <andersson@xxxxxxxxxx> wrote:
> > > >> On Fri, Jul 07, 2023 at 05:27:39PM +0200, Ulf Hansson wrote:
> > > >
> > > > If I understand correctly, you are suggesting that each platform
> > > > maintainer should merge the immutable branch with patch1 - and that I
> > > > should send the patches (based on the immutable branch) to each of the
> > > > platform maintainers to manage. Instead of one pull request with
> > > > everything directly to Arnd, right?
> > > >
> >
> > That's what I suggest, yes.
> >
> > > > This still means that Arnd will have to resolve the conflicts as the
> > > > pull requests arrive in his inbox.
> > > >
> >
> > Yes, we will have N maintainers modifying drivers/genpd/Makefile, all
> > adding their single obj-y line. A quite trivial conflict to manage.
> >
> > > > Although, I guess what you are looking for is less work for the soc/
> > > > maintainers, which seems reasonable. Although, in that case, I might
> > > > as well share an immutable branch with the complete series, rather
> > > > than just patch1. That should help Arnd too, I think.
> > > >
> > > > Again, let's leave the call to Arnd on what to do.
> > >
> > > I think it's much easier for me to pick up a single branch with
> > > all of your patches. For platform maintainers, other changes can
> > > go one of two ways:
> > >
> > > - send a normal pull requests with changes against the same
> > > files, and have me take care of any conflicts where they
> > > arise. Since most of the changes are just simple file moves
> > > rather than changing file contents, 'git mergetool' handles
> > > these fine is most cases
> > >
> > > - If there is a non-obvious merge, the entire genpd branch
> > > can be shared as an immutable branch, with patches for
> > > a particular platform rebased on top of that branch.
> > >
> >
> > I already have a set of patches to these drivers in my tree for v6.6, in
> > their current location.
>
> Right.
>
> >
> > I'm afraid I am unable to see how we're going to handle the merge
> > conflict you're going to create in linux-next. Perhaps you're proposing
> > to just never publish Ulf's patches to linux-next?
> >
> > By me merging the immutable patch 1 and the qcom-patch, the conflicts
> > would be minimal, and except for the genpd/Makefile entirely handled by
> > me.
>
> If you would merge the immutable branch containing the complete
> series, that should work too, rather than just patch1, right?
>

I believe git should sort that out as well. If so I have no concerns
with that option.

Regards,
Bjorn

> As a heads up, I am planning to send the pull request with the v2
> series tomorrow, allowing a few more acks to arrive. At that point I
> will announce the immutable branch too.
>
> Kind regards
> Uffe