From: Bartosz Golaszewski <brgl@bgdev.pl>
To: "Ahmad Fatoum" <a.fatoum@pengutronix.de>,
"Kent Gibson" <warthog618@gmail.com>,
"Jan Lübbe" <jlu@pengutronix.de>, "Marek Vasut" <marex@denx.de>,
"Geert Uytterhoeven" <geert+renesas@glider.be>,
"Linus Walleij" <linus.walleij@linaro.org>,
"Andy Shevchenko" <andriy.shevchenko@intel.com>,
"Bartosz Golaszewski" <brgl@bgdev.pl>
Cc: linux-gpio@vger.kernel.org, linux-kernel@vger.kernel.org,
Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Subject: [PATCH v3 03/10] gpio: sysfs: only get the dirent reference for the value attr once
Date: Mon, 30 Jun 2025 14:37:10 +0200 [thread overview]
Message-ID: <20250630-gpio-sysfs-chip-export-v3-3-b997be9b7137@linaro.org> (raw)
In-Reply-To: <20250630-gpio-sysfs-chip-export-v3-0-b997be9b7137@linaro.org>
From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
There's no reason to retrieve the reference to the sysfs dirent every
time we request an interrupt, we can as well only do it once when
exporting the GPIO. While at it: change the name of the kernfs node
field to value_class_node ahead of adding a second one for the
chip-local GPIO line value attribute.
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
drivers/gpio/gpiolib-sysfs.c | 23 +++++++++++++----------
1 file changed, 13 insertions(+), 10 deletions(-)
diff --git a/drivers/gpio/gpiolib-sysfs.c b/drivers/gpio/gpiolib-sysfs.c
index 990db0405cc86c42bad61295dc823f970199534e..80bb58c96b9b45ce4e1a4def8085d71baa0ce45d 100644
--- a/drivers/gpio/gpiolib-sysfs.c
+++ b/drivers/gpio/gpiolib-sysfs.c
@@ -36,7 +36,7 @@ struct gpiod_data {
struct gpio_desc *desc;
struct mutex mutex;
- struct kernfs_node *value_kn;
+ struct kernfs_node *value_class_node;
int irq;
unsigned char irq_flags;
@@ -156,7 +156,7 @@ static irqreturn_t gpio_sysfs_irq(int irq, void *priv)
{
struct gpiod_data *data = priv;
- sysfs_notify_dirent(data->value_kn);
+ sysfs_notify_dirent(data->value_class_node);
return IRQ_HANDLED;
}
@@ -177,10 +177,6 @@ static int gpio_sysfs_request_irq(struct device *dev, unsigned char flags)
if (data->irq < 0)
return -EIO;
- data->value_kn = sysfs_get_dirent(dev->kobj.sd, "value");
- if (!data->value_kn)
- return -ENODEV;
-
irq_flags = IRQF_SHARED;
if (flags & GPIO_IRQF_TRIGGER_FALLING) {
irq_flags |= test_bit(FLAG_ACTIVE_LOW, &desc->flags) ?
@@ -203,7 +199,7 @@ static int gpio_sysfs_request_irq(struct device *dev, unsigned char flags)
*/
ret = gpiochip_lock_as_irq(guard.gc, gpio_chip_hwgpio(desc));
if (ret < 0)
- goto err_put_kn;
+ goto err_clr_bits;
ret = request_any_context_irq(data->irq, gpio_sysfs_irq, irq_flags,
"gpiolib", data);
@@ -216,10 +212,9 @@ static int gpio_sysfs_request_irq(struct device *dev, unsigned char flags)
err_unlock:
gpiochip_unlock_as_irq(guard.gc, gpio_chip_hwgpio(desc));
-err_put_kn:
+err_clr_bits:
clear_bit(FLAG_EDGE_RISING, &desc->flags);
clear_bit(FLAG_EDGE_FALLING, &desc->flags);
- sysfs_put(data->value_kn);
return ret;
}
@@ -242,7 +237,6 @@ static void gpio_sysfs_free_irq(struct device *dev)
gpiochip_unlock_as_irq(guard.gc, gpio_chip_hwgpio(desc));
clear_bit(FLAG_EDGE_RISING, &desc->flags);
clear_bit(FLAG_EDGE_FALLING, &desc->flags);
- sysfs_put(data->value_kn);
}
static const char *const trigger_names[] = {
@@ -726,8 +720,16 @@ int gpiod_export(struct gpio_desc *desc, bool direction_may_change)
goto err_free_data;
}
+ data->value_class_node = sysfs_get_dirent(dev->kobj.sd, "value");
+ if (!data->value_class_node) {
+ status = -ENODEV;
+ goto err_unregister_device;
+ }
+
return 0;
+err_unregister_device:
+ device_unregister(dev);
err_free_data:
kfree(data);
err_clear_bit:
@@ -804,6 +806,7 @@ void gpiod_unexport(struct gpio_desc *desc)
data = dev_get_drvdata(dev);
clear_bit(FLAG_EXPORT, &desc->flags);
+ sysfs_put(data->value_class_node);
device_unregister(dev);
/*
--
2.48.1
next prev parent reply other threads:[~2025-06-30 12:37 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-06-30 12:37 [PATCH v3 00/10] gpio: sysfs: add a per-chip export/unexport attribute pair Bartosz Golaszewski
2025-06-30 12:37 ` [PATCH v3 01/10] gpio: sysfs: use gpiod_is_equal() to compare GPIO descriptors Bartosz Golaszewski
2025-07-04 8:05 ` Linus Walleij
2025-06-30 12:37 ` [PATCH v3 02/10] gpio: sysfs: add a parallel class device for each GPIO chip using device IDs Bartosz Golaszewski
2025-06-30 12:37 ` Bartosz Golaszewski [this message]
2025-07-04 8:05 ` [PATCH v3 03/10] gpio: sysfs: only get the dirent reference for the value attr once Linus Walleij
2025-06-30 12:37 ` [PATCH v3 04/10] gpio: sysfs: pass gpiod_data directly to internal GPIO sysfs functions Bartosz Golaszewski
2025-06-30 12:37 ` [PATCH v3 05/10] gpio: sysfs: rename the data variable in gpiod_(un)export() Bartosz Golaszewski
2025-06-30 12:37 ` [PATCH v3 06/10] gpio: sysfs: don't use driver data in sysfs callbacks for line attributes Bartosz Golaszewski
2025-06-30 13:27 ` Andy Shevchenko
2025-06-30 13:39 ` Bartosz Golaszewski
2025-06-30 14:22 ` Andy Shevchenko
2025-06-30 12:37 ` [PATCH v3 07/10] gpio: sysfs: don't look up exported lines as class devices Bartosz Golaszewski
2025-06-30 12:37 ` [PATCH v3 08/10] gpio: sysfs: export the GPIO directory locally in the gpiochip<id> directory Bartosz Golaszewski
2025-07-04 8:06 ` Linus Walleij
2025-06-30 12:37 ` [PATCH v3 09/10] gpio: sysfs: allow disabling the legacy parts of the GPIO sysfs interface Bartosz Golaszewski
2025-06-30 12:37 ` [PATCH v3 10/10] gpio: TODO: remove the task for the sysfs rework Bartosz Golaszewski
2025-07-04 8:09 ` Linus Walleij
2025-07-01 14:05 ` [PATCH v3 00/10] gpio: sysfs: add a per-chip export/unexport attribute pair Andy Shevchenko
2025-07-02 3:54 ` Kent Gibson
2025-07-02 9:45 ` Bartosz Golaszewski
2025-07-02 10:12 ` Kent Gibson
2025-07-02 10:28 ` Bartosz Golaszewski
2025-07-02 11:01 ` Kent Gibson
2025-07-02 11:59 ` Andy Shevchenko
2025-07-02 12:42 ` Kent Gibson
2025-07-02 12:05 ` Geert Uytterhoeven
2025-07-02 11:55 ` Andy Shevchenko
2025-07-02 13:18 ` Bartosz Golaszewski
2025-07-04 8:21 ` Linus Walleij
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250630-gpio-sysfs-chip-export-v3-3-b997be9b7137@linaro.org \
--to=brgl@bgdev.pl \
--cc=a.fatoum@pengutronix.de \
--cc=andriy.shevchenko@intel.com \
--cc=bartosz.golaszewski@linaro.org \
--cc=geert+renesas@glider.be \
--cc=jlu@pengutronix.de \
--cc=linus.walleij@linaro.org \
--cc=linux-gpio@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=marex@denx.de \
--cc=warthog618@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).