[PATCH] gpio: 74x164: Enable output pins after registers are reset

[PATCH] gpio: 74x164: Enable output pins after registers are reset

[Arturas Moskvinas]

Move output enabling after chip registers are cleared.

Signed-off-by: Arturas Moskvinas <arturas.moskvinas@gmail.com>
---
 drivers/gpio/gpio-74x164.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/gpio/gpio-74x164.c b/drivers/gpio/gpio-74x164.c
index e00c33310517..753e7be039e4 100644
--- a/drivers/gpio/gpio-74x164.c
+++ b/drivers/gpio/gpio-74x164.c
@@ -127,8 +127,6 @@ static int gen_74x164_probe(struct spi_device *spi)
 	if (IS_ERR(chip->gpiod_oe))
 		return PTR_ERR(chip->gpiod_oe);
 
-	gpiod_set_value_cansleep(chip->gpiod_oe, 1);
-
 	spi_set_drvdata(spi, chip);
 
 	chip->gpio_chip.label = spi->modalias;
@@ -153,6 +151,8 @@ static int gen_74x164_probe(struct spi_device *spi)
 		goto exit_destroy;
 	}
 
+	gpiod_set_value_cansleep(chip->gpiod_oe, 1);
+
 	ret = gpiochip_add_data(&chip->gpio_chip, chip);
 	if (!ret)
 		return 0;

base-commit: d206a76d7d2726f3b096037f2079ce0bd3ba329b
-- 
2.43.2

Re: [PATCH] gpio: 74x164: Enable output pins after registers are reset

[Andy Shevchenko]

On Mon, Feb 26, 2024 at 03:46:56PM +0200, Arturas Moskvinas wrote:
> Move output enabling after chip registers are cleared.

Does this fix anything? If so, maybe elaborate a bit the potential behavioural
changes on the real lines.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] gpio: 74x164: Enable output pins after registers are reset

[Arturas Moskvinas]

Hello,

On 2/26/24 16:01, Andy Shevchenko wrote:
> On Mon, Feb 26, 2024 at 03:46:56PM +0200, Arturas Moskvinas wrote:
>> Move output enabling after chip registers are cleared.
> Does this fix anything? If so, maybe elaborate a bit the potential behavioural
> changes on the real lines.

Chip outputs are enabled[1] before actual reset is performed[2] which 
might cause pin output value to flip flop if previous pin value was set 
to 1 in chip. Change fixes that behavior by making sure chip is fully 
reset before all outputs are enabled.

Flip-flop can be noticed when module is removed and inserted again and 
one of the pins was changed to 1 before removal. 100 microsecond 
flipping is noticeable on oscilloscope (100khz SPI bus).

For a properly reset chip - output is enabled around 100 microseconds 
(on 100khz SPI bus) later during probing process hence should be 
irrelevant behavioral change.

[1] - 
https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L130
[2] - 
https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L150

Arturas Moskvinas

Re: [PATCH] gpio: 74x164: Enable output pins after registers are reset

[Bartosz Golaszewski]

On Tue, Feb 27, 2024 at 7:58 AM Arturas Moskvinas
<arturas.moskvinas@gmail.com> wrote:
>
> Hello,
>
> On 2/26/24 16:01, Andy Shevchenko wrote:
> > On Mon, Feb 26, 2024 at 03:46:56PM +0200, Arturas Moskvinas wrote:
> >> Move output enabling after chip registers are cleared.
> > Does this fix anything? If so, maybe elaborate a bit the potential behavioural
> > changes on the real lines.
>
> Chip outputs are enabled[1] before actual reset is performed[2] which
> might cause pin output value to flip flop if previous pin value was set
> to 1 in chip. Change fixes that behavior by making sure chip is fully
> reset before all outputs are enabled.
>
> Flip-flop can be noticed when module is removed and inserted again and
> one of the pins was changed to 1 before removal. 100 microsecond
> flipping is noticeable on oscilloscope (100khz SPI bus).
>
> For a properly reset chip - output is enabled around 100 microseconds
> (on 100khz SPI bus) later during probing process hence should be
> irrelevant behavioral change.
>
> [1] -
> https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L130
> [2] -
> https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L150
>
> Arturas Moskvinas

And this is precisely the kind of information that needs to go into
commit messages. I can tell *what* you're doing by looking at the
code. What I can't tell is *why*.

Bartosz

Re: [PATCH] gpio: 74x164: Enable output pins after registers are reset

[Andy Shevchenko]

On Tue, Feb 27, 2024 at 02:14:40PM +0100, Bartosz Golaszewski wrote:
> On Tue, Feb 27, 2024 at 7:58 AM Arturas Moskvinas
> <arturas.moskvinas@gmail.com> wrote:
> > On 2/26/24 16:01, Andy Shevchenko wrote:
> > > On Mon, Feb 26, 2024 at 03:46:56PM +0200, Arturas Moskvinas wrote:
> > >> Move output enabling after chip registers are cleared.
> > > Does this fix anything? If so, maybe elaborate a bit the potential behavioural
> > > changes on the real lines.
> >
> > Chip outputs are enabled[1] before actual reset is performed[2] which
> > might cause pin output value to flip flop if previous pin value was set
> > to 1 in chip. Change fixes that behavior by making sure chip is fully
> > reset before all outputs are enabled.
> >
> > Flip-flop can be noticed when module is removed and inserted again and
> > one of the pins was changed to 1 before removal. 100 microsecond
> > flipping is noticeable on oscilloscope (100khz SPI bus).
> >
> > For a properly reset chip - output is enabled around 100 microseconds
> > (on 100khz SPI bus) later during probing process hence should be
> > irrelevant behavioral change.
> >
> > [1] -
> > https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L130
> > [2] -
> > https://elixir.bootlin.com/linux/v6.7.4/source/drivers/gpio/gpio-74x164.c#L150
> >
> > Arturas Moskvinas
> 
> And this is precisely the kind of information that needs to go into
> commit messages. I can tell *what* you're doing by looking at the
> code. What I can't tell is *why*.

+1. Please, add this to the commit message of v2, also try to find the commit
that you can mark to be fixed with help of Fixes tag.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] gpio: 74x164: Enable output pins after registers are reset

[Arturas Moskvinas]

Hello,

...
> > And this is precisely the kind of information that needs to go into
> > commit messages. I can tell *what* you're doing by looking at the
> > code. What I can't tell is *why*.
>
> +1. Please, add this to the commit message of v2, also try to find the commit
> that you can mark to be fixed with help of Fixes tag.

Thanks for suggestion regarding Fixes!

I thought maybe I should as well move whole GPIO initialization[0]
down to the same place I move "gpiod_set_value_cansleep(
chip->gpiod_oe, 1)" in patch v2? I think knowledge that a pin is
brought up later during probing process might be forgotten later, it
will slightly complicate code due to need to clean mutex though.

[0] https://elixir.bootlin.com/linux/v6.7.6/source/drivers/gpio/gpio-74x164.c#L125

Arturas Moskvinas