Re: [PATCH v2 7/7] i2c: designware: Group all DesignWare drivers under a single option

From: Conor Dooley
Date: Mon Sep 09 2024 - 09:05:19 EST


On Mon, Sep 09, 2024 at 02:21:32PM +0200, Andi Shyti wrote:
> Hi Conor,
>
> On Mon, Sep 09, 2024 at 12:22:22PM GMT, Conor Dooley wrote:
> > On Thu, Sep 05, 2024 at 10:46:02PM +0200, Andi Shyti wrote:
> > > Hi Heikki,
> > >
> > > On Tue, Sep 03, 2024 at 05:25:06PM GMT, Heikki Krogerus wrote:
> > > > There are quite a few drivers and options for the DesignWare
> > > > I2C adapter in the Kconfig. Grouping all of them under the
> > > > I2C_DESIGNWARE_CORE. That makes the menuconfig a bit more
> > > > easier to understand.
> > > >
> > > > Signed-off-by: Heikki Krogerus <heikki.krogerus@xxxxxxxxxxxxxxx>
> > >
> > > Thanks for your patch, I can take this only after the other
> > > patches have been taken in.
> >
> > I assume then that you're expecting the prereqs to go through the
> > various arch trees? Is this not trivial enough that you could chuck it
> > on a dedicated branch in your tree and if, for some reason, there's a
> > non-trivial conflict the affected could pull it in?
>
> Thanks for your advice: considering that cross-domain patches are
> difficult to handle, I appreciate any suggestion on how to deal
> with them :-)

It's also helpful if the submitter of a series proposes what path
patches should take upstream. I find that people all look at something
like this and see "someone else's problem" written all over it, and
would be perfectly happy to accept a proposal from a submitter that it
all goes via I2C.

> Are you suggesting to set these up in a different branch and send
> a pull request out of that branch?

What I meant was put them in a dedicated branch (so probably on top of
-rc1 or similar) and merge that branch into your for-next branch. If
someone then was to run into a non-trivial conflict you could tag the
dedicated branch and they could pull it into their tree.

> I don't want here to step into other people branches if they
> don't ack or express any thought.

Sure, I wouldn't either, but it far from abnormal for driver subsystem
maintainer to take these sort of rename/dependency addition patches and
the normal path for 3 of the 5 architectures affected here is via Arnd
and I figure there's a low probability of the maintainers of those
architectures having their feels hurt.

Cheers,
Conor.

Attachment: signature.asc
Description: PGP signature