mbox series

[00/13] gpiolib: add gpiods_set_array_value_cansleep

Message ID 20250131-gpio-set-array-helper-v1-0-991c8ccb4d6e@baylibre.com (mailing list archive)
Headers show
Series gpiolib: add gpiods_set_array_value_cansleep | expand

Message

David Lechner Jan. 31, 2025, 8:24 p.m. UTC
This series was inspired by some minor annoyance I have experienced a
few times in recent reviews.

Calling gpiod_set_array_value_cansleep() can be quite verbose due to
having so many parameters. In most cases, we already have a struct
gpio_descs that contains the first 3 parameters so we end up with 3 (or
often even 6) pointer indirections at each call site. Also, people have
a tendency to want to hard-code the first argument instead of using
struct gpio_descs.ndescs, often without checking that ndescs >= the
hard-coded value.

So I'm proposing that we add a gpiods_set_array_value_cansleep()
function that is a wrapper around gpiod_set_array_value_cansleep()
that has struct gpio_descs as the first parameter to make it a bit
easier to read the code and avoid the hard-coding temptation.

I've just done gpiods_set_array_value_cansleep() for now since there
were over 10 callers of this one. There aren't as many callers of
the get and atomic variants, but we can add those too if this seems
like a useful thing to do.

---
David Lechner (13):
      gpiolib: add gpiods_set_array_value_cansleep()
      auxdisplay: seg-led-gpio: use gpiods_set_array_value_cansleep
      bus: ts-nbus: validate ts,data-gpios array size
      bus: ts-nbus: use gpiods_set_array_value_cansleep
      gpio: max3191x: use gpiods_set_array_value_cansleep
      iio: adc: ad7606: use gpiods_set_array_value_cansleep
      iio: amplifiers: hmc425a: use gpiods_set_array_value_cansleep
      iio: resolver: ad2s1210: use gpiods_set_array_value_cansleep
      mmc: pwrseq_simple: use gpiods_set_array_value_cansleep
      mux: gpio: use gpiods_set_array_value_cansleep
      net: mdio: mux-gpio: use gpiods_set_array_value_cansleep
      phy: mapphone-mdm6600: use gpiods_set_array_value_cansleep
      ASoC: adau1701: use gpiods_set_array_value_cansleep

 drivers/auxdisplay/seg-led-gpio.c           |  3 +--
 drivers/bus/ts-nbus.c                       | 10 ++++++----
 drivers/gpio/gpio-max3191x.c                | 18 +++++++-----------
 drivers/iio/adc/ad7606.c                    |  3 +--
 drivers/iio/adc/ad7606_spi.c                |  3 +--
 drivers/iio/amplifiers/hmc425a.c            |  3 +--
 drivers/iio/resolver/ad2s1210.c             |  8 ++------
 drivers/mmc/core/pwrseq_simple.c            |  3 +--
 drivers/mux/gpio.c                          |  4 +---
 drivers/net/mdio/mdio-mux-gpio.c            |  3 +--
 drivers/phy/motorola/phy-mapphone-mdm6600.c |  4 +---
 include/linux/gpio/consumer.h               |  7 +++++++
 sound/soc/codecs/adau1701.c                 |  4 +---
 13 files changed, 31 insertions(+), 42 deletions(-)
---
base-commit: df4b2bbff898227db0c14264ac7edd634e79f755
change-id: 20250131-gpio-set-array-helper-bd4a328370d3

Best regards,

Comments

Andrew Lunn Jan. 31, 2025, 8:38 p.m. UTC | #1
> So I'm proposing that we add a gpiods_set_array_value_cansleep()
> function that is a wrapper around gpiod_set_array_value_cansleep()
> that has struct gpio_descs as the first parameter to make it a bit
> easier to read the code and avoid the hard-coding temptation.
 
This looks reasonable.

How do you plan to get it merged, since you cross a lot of subsystems
here.

	Andrew
David Lechner Jan. 31, 2025, 8:51 p.m. UTC | #2
On 1/31/25 2:38 PM, Andrew Lunn wrote:
>> So I'm proposing that we add a gpiods_set_array_value_cansleep()
>> function that is a wrapper around gpiod_set_array_value_cansleep()
>> that has struct gpio_descs as the first parameter to make it a bit
>> easier to read the code and avoid the hard-coding temptation.
>  
> This looks reasonable.
> 
> How do you plan to get it merged, since you cross a lot of subsystems
> here.
> 
> 	Andrew

Since these are mostly small changes and most of the touched drivers aren't
seeing much action, I think it would be OK for as much as possible to go through
the GPIO tree.

We might need an immutable branch from that though since I know that iio: adc:
ad7606 is currently being actively worked on.

If there are any patches leftover that don't get acked to go through the GPIO
tree, I can resubmit them after the next kernel release cycle since none of
this is urgent anyway.
Linus Walleij Jan. 31, 2025, 10:45 p.m. UTC | #3
Hi David,

On Fri, Jan 31, 2025 at 9:24 PM David Lechner <dlechner@baylibre.com> wrote:

> This series was inspired by some minor annoyance I have experienced a
> few times in recent reviews.
>
> Calling gpiod_set_array_value_cansleep() can be quite verbose due to
> having so many parameters. In most cases, we already have a struct
> gpio_descs that contains the first 3 parameters so we end up with 3 (or
> often even 6) pointer indirections at each call site. Also, people have
> a tendency to want to hard-code the first argument instead of using
> struct gpio_descs.ndescs, often without checking that ndescs >= the
> hard-coded value.
>
> So I'm proposing that we add a gpiods_set_array_value_cansleep()
> function that is a wrapper around gpiod_set_array_value_cansleep()
> that has struct gpio_descs as the first parameter to make it a bit
> easier to read the code and avoid the hard-coding temptation.
>
> I've just done gpiods_set_array_value_cansleep() for now since there
> were over 10 callers of this one. There aren't as many callers of
> the get and atomic variants, but we can add those too if this seems
> like a useful thing to do.

I like it.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>

for the series.

Yours,
Linus Walleij
Bartosz Golaszewski Feb. 1, 2025, 10:36 a.m. UTC | #4
On Fri, 31 Jan 2025 21:24:40 +0100, David Lechner <dlechner@baylibre.com> said:
> This series was inspired by some minor annoyance I have experienced a
> few times in recent reviews.
>
> Calling gpiod_set_array_value_cansleep() can be quite verbose due to
> having so many parameters. In most cases, we already have a struct
> gpio_descs that contains the first 3 parameters so we end up with 3 (or
> often even 6) pointer indirections at each call site. Also, people have
> a tendency to want to hard-code the first argument instead of using
> struct gpio_descs.ndescs, often without checking that ndescs >= the
> hard-coded value.
>
> So I'm proposing that we add a gpiods_set_array_value_cansleep()
> function that is a wrapper around gpiod_set_array_value_cansleep()
> that has struct gpio_descs as the first parameter to make it a bit
> easier to read the code and avoid the hard-coding temptation.
>
> I've just done gpiods_set_array_value_cansleep() for now since there
> were over 10 callers of this one. There aren't as many callers of
> the get and atomic variants, but we can add those too if this seems
> like a useful thing to do.
>
> ---
> David Lechner (13):
>       gpiolib: add gpiods_set_array_value_cansleep()
>       auxdisplay: seg-led-gpio: use gpiods_set_array_value_cansleep
>       bus: ts-nbus: validate ts,data-gpios array size
>       bus: ts-nbus: use gpiods_set_array_value_cansleep
>       gpio: max3191x: use gpiods_set_array_value_cansleep
>       iio: adc: ad7606: use gpiods_set_array_value_cansleep
>       iio: amplifiers: hmc425a: use gpiods_set_array_value_cansleep
>       iio: resolver: ad2s1210: use gpiods_set_array_value_cansleep
>       mmc: pwrseq_simple: use gpiods_set_array_value_cansleep
>       mux: gpio: use gpiods_set_array_value_cansleep
>       net: mdio: mux-gpio: use gpiods_set_array_value_cansleep
>       phy: mapphone-mdm6600: use gpiods_set_array_value_cansleep
>       ASoC: adau1701: use gpiods_set_array_value_cansleep
>
>  drivers/auxdisplay/seg-led-gpio.c           |  3 +--
>  drivers/bus/ts-nbus.c                       | 10 ++++++----
>  drivers/gpio/gpio-max3191x.c                | 18 +++++++-----------
>  drivers/iio/adc/ad7606.c                    |  3 +--
>  drivers/iio/adc/ad7606_spi.c                |  3 +--
>  drivers/iio/amplifiers/hmc425a.c            |  3 +--
>  drivers/iio/resolver/ad2s1210.c             |  8 ++------
>  drivers/mmc/core/pwrseq_simple.c            |  3 +--
>  drivers/mux/gpio.c                          |  4 +---
>  drivers/net/mdio/mdio-mux-gpio.c            |  3 +--
>  drivers/phy/motorola/phy-mapphone-mdm6600.c |  4 +---
>  include/linux/gpio/consumer.h               |  7 +++++++
>  sound/soc/codecs/adau1701.c                 |  4 +---
>  13 files changed, 31 insertions(+), 42 deletions(-)
> ---
> base-commit: df4b2bbff898227db0c14264ac7edd634e79f755
> change-id: 20250131-gpio-set-array-helper-bd4a328370d3
>
> Best regards,
> --
> David Lechner <dlechner@baylibre.com>
>
>

This looks good to me except for one thing: the function prefix. I would
really appreciate it if we could stay within the existing gpiod_ namespace and
not add a new one in the form of gpiods_.

Maybe: gpiod_multiple_set_ or gpiod_collected_set...?

Bartosz
Jonathan Cameron Feb. 1, 2025, 12:45 p.m. UTC | #5
On Fri, 31 Jan 2025 14:51:52 -0600
David Lechner <dlechner@baylibre.com> wrote:

> On 1/31/25 2:38 PM, Andrew Lunn wrote:
> >> So I'm proposing that we add a gpiods_set_array_value_cansleep()
> >> function that is a wrapper around gpiod_set_array_value_cansleep()
> >> that has struct gpio_descs as the first parameter to make it a bit
> >> easier to read the code and avoid the hard-coding temptation.  
> >  
> > This looks reasonable.
> > 
> > How do you plan to get it merged, since you cross a lot of subsystems
> > here.
> > 
> > 	Andrew  
> 
> Since these are mostly small changes and most of the touched drivers aren't
> seeing much action, I think it would be OK for as much as possible to go through
> the GPIO tree.
> 
> We might need an immutable branch from that though since I know that iio: adc:
> ad7606 is currently being actively worked on.
Looks good to me (subject to requested name change from Bartosz)
I'd suggest an immutable with patch 1 then up to each subsystem
maintainer to pick that up or wait for next cycle.

Always hard to predict what else will get worked on at this stage
of a cycle.

Jonathan


> 
> If there are any patches leftover that don't get acked to go through the GPIO
> tree, I can resubmit them after the next kernel release cycle since none of
> this is urgent anyway.
>
Andy Shevchenko Feb. 1, 2025, 3:10 p.m. UTC | #6
On Sat, Feb 1, 2025 at 12:36 PM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
>
> On Fri, 31 Jan 2025 21:24:40 +0100, David Lechner <dlechner@baylibre.com> said:
> > This series was inspired by some minor annoyance I have experienced a
> > few times in recent reviews.
> >
> > Calling gpiod_set_array_value_cansleep() can be quite verbose due to
> > having so many parameters. In most cases, we already have a struct
> > gpio_descs that contains the first 3 parameters so we end up with 3 (or
> > often even 6) pointer indirections at each call site. Also, people have
> > a tendency to want to hard-code the first argument instead of using
> > struct gpio_descs.ndescs, often without checking that ndescs >= the
> > hard-coded value.
> >
> > So I'm proposing that we add a gpiods_set_array_value_cansleep()
> > function that is a wrapper around gpiod_set_array_value_cansleep()
> > that has struct gpio_descs as the first parameter to make it a bit
> > easier to read the code and avoid the hard-coding temptation.
> >
> > I've just done gpiods_set_array_value_cansleep() for now since there
> > were over 10 callers of this one. There aren't as many callers of
> > the get and atomic variants, but we can add those too if this seems
> > like a useful thing to do.

> This looks good to me except for one thing: the function prefix. I would
> really appreciate it if we could stay within the existing gpiod_ namespace and
> not add a new one in the form of gpiods_.
>
> Maybe: gpiod_multiple_set_ or gpiod_collected_set...?

+1 here, i.e. I like the idea, but the naming needs to be amended.
David Lechner Feb. 1, 2025, 4:09 p.m. UTC | #7
On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:
> On Fri, 31 Jan 2025 21:24:40 +0100, David Lechner <dlechner@baylibre.com> said:
>> This series was inspired by some minor annoyance I have experienced a
>> few times in recent reviews.
>>
>> Calling gpiod_set_array_value_cansleep() can be quite verbose due to
>> having so many parameters. In most cases, we already have a struct
>> gpio_descs that contains the first 3 parameters so we end up with 3 (or
>> often even 6) pointer indirections at each call site. Also, people have
>> a tendency to want to hard-code the first argument instead of using
>> struct gpio_descs.ndescs, often without checking that ndescs >= the
>> hard-coded value.
>>
>> So I'm proposing that we add a gpiods_set_array_value_cansleep()
>> function that is a wrapper around gpiod_set_array_value_cansleep()
>> that has struct gpio_descs as the first parameter to make it a bit
>> easier to read the code and avoid the hard-coding temptation.
>>
>> I've just done gpiods_set_array_value_cansleep() for now since there
>> were over 10 callers of this one. There aren't as many callers of
>> the get and atomic variants, but we can add those too if this seems
>> like a useful thing to do.
>>
>> ---
>> David Lechner (13):
>>       gpiolib: add gpiods_set_array_value_cansleep()
>>       auxdisplay: seg-led-gpio: use gpiods_set_array_value_cansleep
>>       bus: ts-nbus: validate ts,data-gpios array size
>>       bus: ts-nbus: use gpiods_set_array_value_cansleep
>>       gpio: max3191x: use gpiods_set_array_value_cansleep
>>       iio: adc: ad7606: use gpiods_set_array_value_cansleep
>>       iio: amplifiers: hmc425a: use gpiods_set_array_value_cansleep
>>       iio: resolver: ad2s1210: use gpiods_set_array_value_cansleep
>>       mmc: pwrseq_simple: use gpiods_set_array_value_cansleep
>>       mux: gpio: use gpiods_set_array_value_cansleep
>>       net: mdio: mux-gpio: use gpiods_set_array_value_cansleep
>>       phy: mapphone-mdm6600: use gpiods_set_array_value_cansleep
>>       ASoC: adau1701: use gpiods_set_array_value_cansleep
>>
>>  drivers/auxdisplay/seg-led-gpio.c           |  3 +--
>>  drivers/bus/ts-nbus.c                       | 10 ++++++----
>>  drivers/gpio/gpio-max3191x.c                | 18 +++++++-----------
>>  drivers/iio/adc/ad7606.c                    |  3 +--
>>  drivers/iio/adc/ad7606_spi.c                |  3 +--
>>  drivers/iio/amplifiers/hmc425a.c            |  3 +--
>>  drivers/iio/resolver/ad2s1210.c             |  8 ++------
>>  drivers/mmc/core/pwrseq_simple.c            |  3 +--
>>  drivers/mux/gpio.c                          |  4 +---
>>  drivers/net/mdio/mdio-mux-gpio.c            |  3 +--
>>  drivers/phy/motorola/phy-mapphone-mdm6600.c |  4 +---
>>  include/linux/gpio/consumer.h               |  7 +++++++
>>  sound/soc/codecs/adau1701.c                 |  4 +---
>>  13 files changed, 31 insertions(+), 42 deletions(-)
>> ---
>> base-commit: df4b2bbff898227db0c14264ac7edd634e79f755
>> change-id: 20250131-gpio-set-array-helper-bd4a328370d3
>>
>> Best regards,
>> --
>> David Lechner <dlechner@baylibre.com>
>>
>>
> 
> This looks good to me except for one thing: the function prefix. I would
> really appreciate it if we could stay within the existing gpiod_ namespace and
> not add a new one in the form of gpiods_.
> 
> Maybe: gpiod_multiple_set_ or gpiod_collected_set...?
> 
> Bartosz

I was waiting for someone to complain about the naming. ;-)

I was going for as short as possible, but OK, the most obvious prefix to me
would be `gpio_descs_...` (to match the first parameter). Any objections to
that?
Bartosz Golaszewski Feb. 1, 2025, 4:14 p.m. UTC | #8
On Sat, Feb 1, 2025 at 5:09 PM David Lechner <dlechner@baylibre.com> wrote:
>
> On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:
> >
> > This looks good to me except for one thing: the function prefix. I would
> > really appreciate it if we could stay within the existing gpiod_ namespace and
> > not add a new one in the form of gpiods_.
> >
> > Maybe: gpiod_multiple_set_ or gpiod_collected_set...?
> >
> > Bartosz
>
> I was waiting for someone to complain about the naming. ;-)
>
> I was going for as short as possible, but OK, the most obvious prefix to me
> would be `gpio_descs_...` (to match the first parameter). Any objections to
> that?
>

Yes, objection! As far as any exported interfaces go: in my book
"gpio_" is the prefix for legacy symbols we want to go away and
"gpiod_" is the prefix for current, descriptor-based API. Anything
else is a no-go. I prefer a longer name that starts with gpiod_ over
anything that's shorter but doesn't.

Bartosz
David Lechner Feb. 1, 2025, 4:17 p.m. UTC | #9
On 2/1/25 10:14 AM, Bartosz Golaszewski wrote:
> On Sat, Feb 1, 2025 at 5:09 PM David Lechner <dlechner@baylibre.com> wrote:
>>
>> On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:
>>>
>>> This looks good to me except for one thing: the function prefix. I would
>>> really appreciate it if we could stay within the existing gpiod_ namespace and
>>> not add a new one in the form of gpiods_.
>>>
>>> Maybe: gpiod_multiple_set_ or gpiod_collected_set...?
>>>
>>> Bartosz
>>
>> I was waiting for someone to complain about the naming. ;-)
>>
>> I was going for as short as possible, but OK, the most obvious prefix to me
>> would be `gpio_descs_...` (to match the first parameter). Any objections to
>> that?
>>
> 
> Yes, objection! As far as any exported interfaces go: in my book
> "gpio_" is the prefix for legacy symbols we want to go away and
> "gpiod_" is the prefix for current, descriptor-based API. Anything
> else is a no-go. I prefer a longer name that starts with gpiod_ over
> anything that's shorter but doesn't.
> 
> Bartosz

Oops, that was a typo. I meant to write gpiod_descs_.
Bartosz Golaszewski Feb. 1, 2025, 4:22 p.m. UTC | #10
On Sat, Feb 1, 2025 at 5:17 PM David Lechner <dlechner@baylibre.com> wrote:
>
> On 2/1/25 10:14 AM, Bartosz Golaszewski wrote:
> > On Sat, Feb 1, 2025 at 5:09 PM David Lechner <dlechner@baylibre.com> wrote:
> >>
> >> On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:
> >>>
> >>> This looks good to me except for one thing: the function prefix. I would
> >>> really appreciate it if we could stay within the existing gpiod_ namespace and
> >>> not add a new one in the form of gpiods_.
> >>>
> >>> Maybe: gpiod_multiple_set_ or gpiod_collected_set...?
> >>>
> >>> Bartosz
> >>
> >> I was waiting for someone to complain about the naming. ;-)
> >>
> >> I was going for as short as possible, but OK, the most obvious prefix to me
> >> would be `gpio_descs_...` (to match the first parameter). Any objections to
> >> that?
> >>
> >
> > Yes, objection! As far as any exported interfaces go: in my book
> > "gpio_" is the prefix for legacy symbols we want to go away and
> > "gpiod_" is the prefix for current, descriptor-based API. Anything
> > else is a no-go. I prefer a longer name that starts with gpiod_ over
> > anything that's shorter but doesn't.
> >
> > Bartosz
>
> Oops, that was a typo. I meant to write gpiod_descs_.

Eh... the D in gpioD already stands for "GPIO Descriptor" but if
there's no better option in your opinion than I guess I can live with
that.

Bart
Andy Shevchenko Feb. 1, 2025, 7:47 p.m. UTC | #11
On Sat, Feb 1, 2025 at 6:22 PM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
> On Sat, Feb 1, 2025 at 5:17 PM David Lechner <dlechner@baylibre.com> wrote:
> > On 2/1/25 10:14 AM, Bartosz Golaszewski wrote:
> > > On Sat, Feb 1, 2025 at 5:09 PM David Lechner <dlechner@baylibre.com> wrote:
> > >> On 2/1/25 4:36 AM, Bartosz Golaszewski wrote:

...

> > >>> This looks good to me except for one thing: the function prefix. I would
> > >>> really appreciate it if we could stay within the existing gpiod_ namespace and
> > >>> not add a new one in the form of gpiods_.
> > >>>
> > >>> Maybe: gpiod_multiple_set_ or gpiod_collected_set...?
> > >>
> > >> I was waiting for someone to complain about the naming. ;-)
> > >>
> > >> I was going for as short as possible, but OK, the most obvious prefix to me
> > >> would be `gpio_descs_...` (to match the first parameter). Any objections to
> > >> that?
> > >
> > > Yes, objection! As far as any exported interfaces go: in my book
> > > "gpio_" is the prefix for legacy symbols we want to go away and
> > > "gpiod_" is the prefix for current, descriptor-based API. Anything
> > > else is a no-go. I prefer a longer name that starts with gpiod_ over
> > > anything that's shorter but doesn't.
> >
> > Oops, that was a typo. I meant to write gpiod_descs_.
>
> Eh... the D in gpioD already stands for "GPIO Descriptor" but if
> there's no better option in your opinion than I guess I can live with
> that.

gpiod_set_many_value_cansleep() ?