[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <fe44039a-4fa9-dab3-cd14-04967b729158@linux.intel.com>
Date: Wed, 27 May 2020 19:31:53 -0500
From: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To: Michael Walle <michael@...le.cc>
Cc: linux-kernel@...r.kernel.org, linux-gpio@...r.kernel.org,
Linus Walleij <linus.walleij@...aro.org>,
Bartosz Golaszewski <bgolaszewski@...libre.com>,
Mark Brown <broonie@...nel.org>,
Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Subject: Re: [PATCH v4 2/2] gpio: add a reusable generic gpio_chip using
regmap
Hi Michael,
>>> +struct gpio_regmap_config {
>>> + struct device *parent;
>>> + struct regmap *regmap;
>>> +
>>> + const char *label;
>>> + int ngpio;
>>
>> could we add a .names field for the gpio_chip, I found this useful for
>> PCM512x GPIO support, e.g.
>
> Sure, I have the names in the device tree.
>
> But I'd prefer that you'd do a patch on top of this (assuming it is
> applied soon), because you can actually test it and there might be
> missing more.
I am happy to report that this gpio-regmap worked like a charm for me,
after I applied the minor diff below (complete code at
https://github.com/plbossart/sound/tree/fix/regmap-gpios).
I worked around my previous comments by forcing the GPIO internal
routing directly in regmap, and that allowed me to only play with the
_set and _dir bases. I see the LEDs and clock selected as before, quite
nice indeed.
The chip->label test is probably wrong, since the gpio_chip structure is
zeroed out if(!chip->label) is always true so the label is always set to
the device name. I don't know what the intent was so just removed that
test - maybe the correct test should be if (!config->label) ?
I added the names support as well, and btw I don't understand how one
would get them through device tree?
I still have a series of odd warnings I didn't have before:
[ 101.400263] WARNING: CPU: 3 PID: 1129 at drivers/gpio/gpiolib.c:4084
gpiod_set_value+0x3f/0x50
This seems to come from
/* Should be using gpiod_set_value_cansleep() */
WARN_ON(desc->gdev->chip->can_sleep);
so maybe we need an option here as well? Or use a different function?
Anyways, that gpio-regmap does simplify my code a great deal so thanks
for this work, much appreciated.
-Pierre
diff --git a/drivers/gpio/gpio-regmap.c b/drivers/gpio/gpio-regmap.c
index 3cb0e8493835..678d644a0a4b 100644
--- a/drivers/gpio/gpio-regmap.c
+++ b/drivers/gpio/gpio-regmap.c
@@ -251,10 +251,8 @@ struct gpio_regmap *gpio_regmap_register(const
struct gpio_regmap_config *config
chip->ngpio = config->ngpio;
chip->can_sleep = true;
- if (!chip->label)
- chip->label = dev_name(config->parent);
- else
- chip->label = config->label;
+ chip->label = config->label;
+ chip->names = config->names;
chip->get = gpio_regmap_get;
if (gpio->reg_set_base && gpio->reg_clr_base)
diff --git a/include/linux/gpio-regmap.h b/include/linux/gpio-regmap.h
index bbdb2d79ef8f..c1f3e36ebf33 100644
--- a/include/linux/gpio-regmap.h
+++ b/include/linux/gpio-regmap.h
@@ -16,6 +16,7 @@ struct gpio_regmap;
* given, the name of the device is used
* @label: (Optional) Descriptive name for GPIO controller.
* If not given, the name of the device is used.
+ * @names: (Optional) Array of names for gpios
* @ngpio: Number of GPIOs
* @reg_dat_base: (Optional) (in) register base address
* @reg_set_base: (Optional) set register base address
@@ -43,6 +44,7 @@ struct gpio_regmap_config {
struct regmap *regmap;
const char *label;
+ const char *const *names;
int ngpio;
unsigned int reg_dat_base;
Powered by blists - more mailing lists