diff mbox series

[4/6] gpio: pisosr: Use devm_gpiochip_add_data() to simplify remove path

Message ID 20230307165432.25484-4-afd@ti.com
State New
Headers
Series [1/6] gpio: ich: Use devm_gpiochip_add_data() to simplify remove path |

Commit Message

Andrew Davis March 7, 2023, 4:54 p.m. UTC 
  Use devm version of gpiochip add function to handle removal for us.

While here update copyright and module author.

Signed-off-by: Andrew Davis <afd@ti.com>
---
 drivers/gpio/gpio-pisosr.c | 20 ++++----------------
 1 file changed, 4 insertions(+), 16 deletions(-)

Comments

andy@kernel.org March 7, 2023, 5:44 p.m. UTC | #1 
On Tue, Mar 07, 2023 at 10:54:30AM -0600, Andrew Davis wrote:
> Use devm version of gpiochip add function to handle removal for us.
> 
> While here update copyright and module author.

...

> -	mutex_destroy(&gpio->lock);

You need to wrap this into devm.
Andrew Davis March 7, 2023, 5:55 p.m. UTC | #2 
On 3/7/23 11:44 AM, Andy Shevchenko wrote:
> On Tue, Mar 07, 2023 at 10:54:30AM -0600, Andrew Davis wrote:
>> Use devm version of gpiochip add function to handle removal for us.
>>
>> While here update copyright and module author.
> 
> ...
> 
>> -	mutex_destroy(&gpio->lock);
> 
> You need to wrap this into devm.
> 

I was thinking that but it seems there is no such thing. Most drivers
just ignore unwinding mutex_init() since it doesn't allocate anything.

mutex_destroy() is a NOP unless you are doing DEBUG builds were
it sets a magic value to check for use-after-free issues.

Andrew
andy@kernel.org March 7, 2023, 6:05 p.m. UTC | #3 
On Tue, Mar 07, 2023 at 11:55:11AM -0600, Andrew Davis wrote:
> On 3/7/23 11:44 AM, Andy Shevchenko wrote:
> > On Tue, Mar 07, 2023 at 10:54:30AM -0600, Andrew Davis wrote:
> > > Use devm version of gpiochip add function to handle removal for us.
> > > 
> > > While here update copyright and module author.

...

> > > -	mutex_destroy(&gpio->lock);
> > 
> > You need to wrap this into devm.
> 
> I was thinking that but it seems there is no such thing. Most drivers
> just ignore unwinding mutex_init() since it doesn't allocate anything.
> 
> mutex_destroy() is a NOP unless you are doing DEBUG builds were
> it sets a magic value to check for use-after-free issues.

In any case it's correct to destroy it.

See, how it's done, for example, here a82c7cf803d9 ("leds: is31fl319x: Wrap
mutex_destroy() for devm_add_action_or_rest()").
Bartosz Golaszewski March 8, 2023, 10:27 a.m. UTC | #4 
On Tue, Mar 7, 2023 at 6:55 PM Andrew Davis <afd@ti.com> wrote:
>
> On 3/7/23 11:44 AM, Andy Shevchenko wrote:
> > On Tue, Mar 07, 2023 at 10:54:30AM -0600, Andrew Davis wrote:
> >> Use devm version of gpiochip add function to handle removal for us.
> >>
> >> While here update copyright and module author.
> >
> > ...
> >
> >> -    mutex_destroy(&gpio->lock);
> >
> > You need to wrap this into devm.
> >
>
> I was thinking that but it seems there is no such thing. Most drivers
> just ignore unwinding mutex_init() since it doesn't allocate anything.
>
> mutex_destroy() is a NOP unless you are doing DEBUG builds were
> it sets a magic value to check for use-after-free issues.
>
> Andrew

And this is precisely why it's useful to destroy it. :)

Bart
diff mbox series

Patch

diff --git a/drivers/gpio/gpio-pisosr.c b/drivers/gpio/gpio-pisosr.c
index 67071bea08c2..eaf65606035d 100644
--- a/drivers/gpio/gpio-pisosr.c
+++ b/drivers/gpio/gpio-pisosr.c
@@ -1,7 +1,7 @@ 
 // SPDX-License-Identifier: GPL-2.0-only
 /*
- * Copyright (C) 2015 Texas Instruments Incorporated - http://www.ti.com/
- *	Andrew F. Davis <afd@ti.com>
+ * Copyright (C) 2015-2023 Texas Instruments Incorporated - https://www.ti.com/
+ *	Andrew Davis <afd@ti.com>
  */
 
 #include <linux/bitmap.h>
@@ -126,8 +126,6 @@  static int pisosr_gpio_probe(struct spi_device *spi)
 	if (!gpio)
 		return -ENOMEM;
 
-	spi_set_drvdata(spi, gpio);
-
 	gpio->chip = template_chip;
 	gpio->chip.parent = dev;
 	of_property_read_u16(dev->of_node, "ngpios", &gpio->chip.ngpio);
@@ -146,7 +144,7 @@  static int pisosr_gpio_probe(struct spi_device *spi)
 
 	mutex_init(&gpio->lock);
 
-	ret = gpiochip_add_data(&gpio->chip, gpio);
+	ret = devm_gpiochip_add_data(dev, &gpio->chip, gpio);
 	if (ret < 0) {
 		dev_err(dev, "Unable to register gpiochip\n");
 		return ret;
@@ -155,15 +153,6 @@  static int pisosr_gpio_probe(struct spi_device *spi)
 	return 0;
 }
 
-static void pisosr_gpio_remove(struct spi_device *spi)
-{
-	struct pisosr_gpio *gpio = spi_get_drvdata(spi);
-
-	gpiochip_remove(&gpio->chip);
-
-	mutex_destroy(&gpio->lock);
-}
-
 static const struct spi_device_id pisosr_gpio_id_table[] = {
 	{ "pisosr-gpio", },
 	{ /* sentinel */ }
@@ -182,11 +171,10 @@  static struct spi_driver pisosr_gpio_driver = {
 		.of_match_table = pisosr_gpio_of_match_table,
 	},
 	.probe = pisosr_gpio_probe,
-	.remove = pisosr_gpio_remove,
 	.id_table = pisosr_gpio_id_table,
 };
 module_spi_driver(pisosr_gpio_driver);
 
-MODULE_AUTHOR("Andrew F. Davis <afd@ti.com>");
+MODULE_AUTHOR("Andrew Davis <afd@ti.com>");
 MODULE_DESCRIPTION("SPI Compatible PISO Shift Register GPIO Driver");
 MODULE_LICENSE("GPL v2");