Re: [RFC v7 4/6] gpio: mpfs: add polarfire soc gpio support

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1616 bytes --]

On Wed, Aug 07, 2024 at 06:55:58PM +0200, Linus Walleij wrote:
> On Tue, Aug 6, 2024 at 7:18 PM Conor Dooley <conor@kernel.org> wrote:
> > On Mon, Aug 05, 2024 at 10:04:53AM +0200, Linus Walleij wrote:
> 
> > > Is it possible to use augmented generic MMIO, i.e just override these
> > > two functions that
> > > need special handling?
> >
> > I'll look into it - as I mentioned under the --- line, I really didn't
> > touch most of the driver and there's comments from Lewis' submission
> > that still apply.
> 
> Thanks Conor, thanks for taking this over, too many patch sets fall
> on the floor.

Meh, it always bugged me that upstreaming driver was given up on because
fixing the interrupts was "hard". It'd be a poor example to others if I
le the driver sit downstream as a result.

> I'm mostly fine merging it like this and then improving
> it in-tree as well, I'm not as insistent on things being perfect before
> merging as long as they are maintained.

The gpio side of things might not be too bad, but the irqchip side is
crap (it has an of_iomap() in it for example*), and if the irqchip driver
needs work it feels sensible to improve on both before merging anything.
Unless of course, you think it would be reasonable to rip the interrupt
support out of the gpio driver, merge that and incrementally add it
while also improving the things you and the earlier review mentioned
w.r.t. regmap?

Cheers,
Conor.

* Getting rid of the of_iomap() needs me to rewrite parts of the clock
driver, which seemed overkill until I knew whether or not the approach
was permitted.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]