GPIO TODO

GPIO TODO

[Peng Fan]

Hi All,

I did two patches to drop of_gpio.h for 
drivers/spi/spi-stm32-ospi.c and drivers/regulator/s5m8767.c.
But when I search lore.kernel.org, I see Andy did some
work on the cleanup, hope there is no duplication with your
work, and I wonder do you plan to cleanup the remaining
ones under drivers/net/ drivers/nfc sound/ ?

Besides the of_gpio.h, anyone working on immutable
irq_chip?

Thanks,
Peng

Re: GPIO TODO

[Andy Shevchenko]

On Tue, Mar 18, 2025 at 07:37:59AM +0000, Peng Fan wrote:
> Hi All,
> 
> I did two patches to drop of_gpio.h for 
> drivers/spi/spi-stm32-ospi.c and drivers/regulator/s5m8767.c.

Dropping of_gpio.h from the latter is not just that, one needs to
convert it to use GPIO descriptor APIs first.

> But when I search lore.kernel.org, I see Andy did some
> work on the cleanup, hope there is no duplication with your
> work, and I wonder do you plan to cleanup the remaining
> ones under drivers/net/ drivers/nfc sound/ ?

There is no duplication for sure, any help will be appreciated.

> Besides the of_gpio.h, anyone working on immutable irq_chip?

As far as I know that work has been staled. Again, it would be nice
to continue.

-- 
With Best Regards,
Andy Shevchenko

RE: GPIO TODO

[Peng Fan]

Hi Andy,

> Subject: Re: GPIO TODO
> 
> On Tue, Mar 18, 2025 at 07:37:59AM +0000, Peng Fan wrote:
> > Hi All,
> >
> > I did two patches to drop of_gpio.h for drivers/spi/spi-stm32-ospi.c
> > and drivers/regulator/s5m8767.c.
> 
> Dropping of_gpio.h from the latter is not just that, one needs to
> convert it to use GPIO descriptor APIs first.

Yes, I did that. Not have device to test, just my best practice to
do the convertion.
https://lore.kernel.org/all/20250318052709.1731747-1-peng.fan@oss.nxp.com/

> 
> > But when I search lore.kernel.org, I see Andy did some work on the
> > cleanup, hope there is no duplication with your work, and I wonder
> do
> > you plan to cleanup the remaining ones under drivers/net/
> drivers/nfc
> > sound/ ?
> 
> There is no duplication for sure, any help will be appreciated.

My pleasure, just have spare time on various stuff.

> 
> > Besides the of_gpio.h, anyone working on immutable irq_chip?
> 
> As far as I know that work has been staled. Again, it would be nice to
> continue.

I will give a look.

Thanks,
Peng.

> 
> --
> With Best Regards,
> Andy Shevchenko
> 

Re: GPIO TODO

[Linus Walleij]

On Tue, Mar 18, 2025 at 8:38 AM Peng Fan <peng.fan@nxp.com> wrote:

> Besides the of_gpio.h, anyone working on immutable
> irq_chip?

This work was initiated by Marc Zyngier and he has since had
to resign irqchip maintenance so it lands with Tomas Gleixner that
has too much to do (I think), but rest assured that he will be
grateful if you pick it up, so just send patches.

Yours,
Linus Walleij

Re: GPIO TODO

[Andy Shevchenko]

On Tue, Mar 18, 2025 at 09:58:43AM +0000, Peng Fan wrote:
> > On Tue, Mar 18, 2025 at 07:37:59AM +0000, Peng Fan wrote:
> > >
> > > I did two patches to drop of_gpio.h for drivers/spi/spi-stm32-ospi.c
> > > and drivers/regulator/s5m8767.c.
> > 
> > Dropping of_gpio.h from the latter is not just that, one needs to
> > convert it to use GPIO descriptor APIs first.
> 
> Yes, I did that. Not have device to test, just my best practice to
> do the convertion.
> https://lore.kernel.org/all/20250318052709.1731747-1-peng.fan@oss.nxp.com/

As far as I can tell this needs a bit more work and also have you checked
the existing DTS files? What about polarity?

> > > But when I search lore.kernel.org, I see Andy did some work on the
> > > cleanup, hope there is no duplication with your work, and I wonder
> > do
> > > you plan to cleanup the remaining ones under drivers/net/
> > drivers/nfc
> > > sound/ ?
> > 
> > There is no duplication for sure, any help will be appreciated.
> 
> My pleasure, just have spare time on various stuff.
> 
> > > Besides the of_gpio.h, anyone working on immutable irq_chip?
> > 
> > As far as I know that work has been staled. Again, it would be nice to
> > continue.
> 
> I will give a look.

-- 
With Best Regards,
Andy Shevchenko

Re: GPIO TODO

[Marc Zyngier]

On Tue, 18 Mar 2025 10:00:16 +0000,
Linus Walleij <linus.walleij@linaro.org> wrote:
> 
> On Tue, Mar 18, 2025 at 8:38 AM Peng Fan <peng.fan@nxp.com> wrote:
> 
> > Besides the of_gpio.h, anyone working on immutable
> > irq_chip?
> 
> This work was initiated by Marc Zyngier and he has since had
> to resign irqchip maintenance so it lands with Tomas Gleixner that
> has too much to do (I think), but rest assured that he will be
> grateful if you pick it up, so just send patches.

You of course still have the option to simply delete anything that is
not yet immutable, or remove the irq_chip patching support. The kernel
has been screaming for three years now, and these drivers are pretty
much unmaintained.

My bet is that nobody will notice, and I suspect tglx will gladly take
patches deleting unmaintained code. Once that's done, you can drop the
immutable flag, which won't serve any purpose anymore.

	M.

-- 
Without deviation from the norm, progress is not possible.

RE: GPIO TODO

[Peng Fan]

Hi Marc, Linus

> Subject: Re: GPIO TODO

Thanks for quick response.

> 
> On Tue, 18 Mar 2025 10:00:16 +0000,
> Linus Walleij <linus.walleij@linaro.org> wrote:
> >
> > On Tue, Mar 18, 2025 at 8:38 AM Peng Fan <peng.fan@nxp.com>
> wrote:
> >
> > > Besides the of_gpio.h, anyone working on immutable irq_chip?
> >
> > This work was initiated by Marc Zyngier and he has since had to
> resign
> > irqchip maintenance so it lands with Tomas Gleixner that has too
> much
> > to do (I think), but rest assured that he will be grateful if you pick
> > it up, so just send patches.

Thanks. I will give a look and see what I could do here.

> 
> You of course still have the option to simply delete anything that is not
> yet immutable, or remove the irq_chip patching support. The kernel
> has been screaming for three years now, and these drivers are pretty
> much unmaintained.

I will do my best to add IRQCHIP_IMMUTABLE support
for remaining irq_chips created by gpio drivers.

Removing the irq_chips from the gpio drivers that not-converted
immutable irq chip(if you mean this) would be rejected, I think.

Thanks,
Peng. 

> 
> My bet is that nobody will notice, and I suspect tglx will gladly take
> patches deleting unmaintained code. Once that's done, you can drop
> the immutable flag, which won't serve any purpose anymore.
> 
> 	M.
> 
> --
> Without deviation from the norm, progress is not possible.

Re: GPIO TODO

[Marc Zyngier]

On Tue, 18 Mar 2025 14:51:40 +0000,
Peng Fan <peng.fan@nxp.com> wrote:
> 
> Hi Marc, Linus
> 
> > Subject: Re: GPIO TODO
> 
> Thanks for quick response.
> 
> > 
> > On Tue, 18 Mar 2025 10:00:16 +0000,
> > Linus Walleij <linus.walleij@linaro.org> wrote:
> > >
> > > On Tue, Mar 18, 2025 at 8:38 AM Peng Fan <peng.fan@nxp.com>
> > wrote:
> > >
> > > > Besides the of_gpio.h, anyone working on immutable irq_chip?
> > >
> > > This work was initiated by Marc Zyngier and he has since had to
> > resign
> > > irqchip maintenance so it lands with Tomas Gleixner that has too
> > much
> > > to do (I think), but rest assured that he will be grateful if you pick
> > > it up, so just send patches.
> 
> Thanks. I will give a look and see what I could do here.
> 
> > 
> > You of course still have the option to simply delete anything that is not
> > yet immutable, or remove the irq_chip patching support. The kernel
> > has been screaming for three years now, and these drivers are pretty
> > much unmaintained.
> 
> I will do my best to add IRQCHIP_IMMUTABLE support
> for remaining irq_chips created by gpio drivers.
> 
> Removing the irq_chips from the gpio drivers that not-converted
> immutable irq chip(if you mean this) would be rejected, I think.

I think it would actually be *very* welcome by quite a few people.

Unmaintained code is a liability, and these GPIO drivers (or at least
their irqchip part) are obviously not looked after in any meaningful
way. Such drivers are userspace facing, and I'm not convinced we want
more unmaintained code in this position.

If you plan to fix these drivers, great. You will be the de-facto
maintainer for them. But it is equally important to evaluate whether
they still have a purpose in the kernel tree.

	M.

-- 
Without deviation from the norm, progress is not possible.

Re: GPIO TODO

[Linus Walleij]

On Tue, Mar 18, 2025 at 3:51 PM Peng Fan <peng.fan@nxp.com> wrote:

> I will do my best to add IRQCHIP_IMMUTABLE support
> for remaining irq_chips created by gpio drivers.

Sweet, I have fixed all the easy ones (I think).
Those that remain probably require a one-by-one approach,
and some tinkering.

Yours,
Linus Walleij