Re: [PATCH] greybus: gpio: switch GPIO portions to use GPIOLIB_IRQCHIP

[Nishad Kamdar]

On Mon, Nov 12, 2018 at 04:15:09PM +0100, Johan Hovold wrote:
> On Fri, Nov 09, 2018 at 01:17:41PM +0530, Nishad Kamdar wrote:
> > Convert the GPIO driver to use the GPIO irqchip library
> > GPIOLIB_IRQCHIP instead of reimplementing the same.
> 
> Thanks for picking this up. Linus W took a stab at it a couple of years
> ago, but it was never completed. You may want to take a look at that
> thread:
> 
> 	https://lkml.kernel.org/r/1476054589-28422-1-git-send-email-linus.walleij@xxxxxxxxxx

Thanks for the reference.

> 
> > Signed-off-by: Nishad Kamdar <nishadkamdar@xxxxxxxxx>
> > ---
> >  drivers/staging/greybus/Kconfig |   1 +
> >  drivers/staging/greybus/gpio.c  | 123 ++++++--------------------------
> >  2 files changed, 21 insertions(+), 103 deletions(-)
> > 
> > diff --git a/drivers/staging/greybus/Kconfig b/drivers/staging/greybus/Kconfig
> > index ab096bcef98c..b571e4e8060b 100644
> > --- a/drivers/staging/greybus/Kconfig
> > +++ b/drivers/staging/greybus/Kconfig
> > @@ -148,6 +148,7 @@ if GREYBUS_BRIDGED_PHY
> >  config GREYBUS_GPIO
> >  	tristate "Greybus GPIO Bridged PHY driver"
> >  	depends on GPIOLIB
> > +	select GPIOLIB_IRQCHIP
> >  	---help---
> >  	  Select this option if you have a device that follows the
> >  	  Greybus GPIO Bridged PHY Class specification.
> > diff --git a/drivers/staging/greybus/gpio.c b/drivers/staging/greybus/gpio.c
> > index b1d4698019a1..32c228bad33a 100644
> > --- a/drivers/staging/greybus/gpio.c
> > +++ b/drivers/staging/greybus/gpio.c
> > @@ -9,9 +9,7 @@
> >  #include <linux/kernel.h>
> >  #include <linux/module.h>
> >  #include <linux/slab.h>
> > -#include <linux/gpio.h>
> > -#include <linux/irq.h>
> 
> I think you should keep irq.h even if the gpio header currently provides
> it.
> 
> > -#include <linux/irqdomain.h>
> 
> Similarly for this one, if you still rely on it.
>

Ok i'll do that.

> > +#include <linux/gpio/driver.h>
> >  #include <linux/mutex.h>
> >  
> >  #include "greybus.h"
> > @@ -40,8 +38,6 @@ struct gb_gpio_controller {
> >  	struct gpio_chip	chip;
> >  	struct irq_chip		irqc;
> >  	struct irq_chip		*irqchip;
> 
> This should not be needed any more either.
>

Just to confirm, by thism you mean only 
struct irq_chip	*irqchip; right?

> > -	struct irq_domain	*irqdomain;
> > -	unsigned int		irq_base;
> >  	irq_flow_handler_t	irq_handler;
> >  	unsigned int		irq_default_type;
> 
> Neither should these two.
> 

Ok.

> >  	struct mutex		irq_lock;
> > @@ -365,6 +361,7 @@ static int gb_gpio_request_handler(struct gb_operation *op)
> >  {
> >  	struct gb_connection *connection = op->connection;
> >  	struct gb_gpio_controller *ggc = gb_connection_get_data(connection);
> > +	struct gpio_chip *gc = &ggc->chip;
> 
> Please name the pointer 'chip' as in the rest of the driver to avoid
> confusion with 'ggc'. But looks like you don't need it at all.
>

Yes.

> >  	struct device *dev = &ggc->gbphy_dev->dev;
> >  	struct gb_message *request;
> >  	struct gb_gpio_irq_event_request *event;
> > @@ -391,7 +388,7 @@ static int gb_gpio_request_handler(struct gb_operation *op)
> >  		return -EINVAL;
> >  	}
> >  
> > -	irq = irq_find_mapping(ggc->irqdomain, event->which);
> > +	irq = irq_find_mapping(gc->irq.domain, event->which);
> >  	if (!irq) {
> >  		dev_err(dev, "failed to find IRQ\n");
> >  		return -EINVAL;
> > @@ -506,68 +503,6 @@ static int gb_gpio_controller_setup(struct gb_gpio_controller *ggc)
> >  	return ret;
> >  }
> 
> > -/**
> > - * gb_gpio_irqchip_remove() - removes an irqchip added to a gb_gpio_controller
> > - * @ggc: the gb_gpio_controller to remove the irqchip from
> > - *
> > - * This is called only from gb_gpio_remove()
> > - */
> > -static void gb_gpio_irqchip_remove(struct gb_gpio_controller *ggc)
> > -{
> > -	unsigned int offset;
> > -
> > -	/* Remove all IRQ mappings and delete the domain */
> > -	if (ggc->irqdomain) {
> > -		for (offset = 0; offset < (ggc->line_max + 1); offset++)
> > -			irq_dispose_mapping(irq_find_mapping(ggc->irqdomain,
> > -							     offset));
> > -		irq_domain_remove(ggc->irqdomain);
> > -	}
> > -
> > -	if (ggc->irqchip)
> > -		ggc->irqchip = NULL;
> > -}
> > -
> >  /**
> >   * gb_gpio_irqchip_add() - adds an irqchip to a gpio chip
> >   * @chip: the gpio chip to add the irqchip to
> > @@ -595,8 +530,7 @@ static int gb_gpio_irqchip_add(struct gpio_chip *chip,
> >  			 unsigned int type)
> >  {
> >  	struct gb_gpio_controller *ggc;
> > -	unsigned int offset;
> > -	unsigned int irq_base;
> > +	unsigned int err;
> >  
> >  	if (!chip || !irqchip)
> >  		return -EINVAL;
> > @@ -606,35 +540,21 @@ static int gb_gpio_irqchip_add(struct gpio_chip *chip,
> >  	ggc->irqchip = irqchip;
> >  	ggc->irq_handler = handler;
> >  	ggc->irq_default_type = type;
> > -	ggc->irqdomain = irq_domain_add_simple(NULL,
> > -					ggc->line_max + 1, first_irq,
> > -					&gb_gpio_domain_ops, chip);
> > -	if (!ggc->irqdomain) {
> > -		ggc->irqchip = NULL;
> > -		return -EINVAL;
> > -	}
> >  
> > -	/*
> > -	 * Prepare the mapping since the irqchip shall be orthogonal to
> > -	 * any gpio calls. If the first_irq was zero, this is
> > -	 * necessary to allocate descriptors for all IRQs.
> > -	 */
> > -	for (offset = 0; offset < (ggc->line_max + 1); offset++) {
> > -		irq_base = irq_create_mapping(ggc->irqdomain, offset);
> > -		if (offset == 0)
> > -			ggc->irq_base = irq_base;
> > +	err = gpiochip_irqchip_add(chip,
> > +				   irqchip,
> > +				   first_irq,
> > +				   ggc->irq_handler,
> > +				   type
> 
> Don't break the line here.
> 

Ok.

> > +				   );
> > +	if (err) {
> > +		ggc->irqchip = NULL;
> > +		return err;
> >  	}
> >  
> >  	return 0;
> >  }
> 
> Drop this function as well and call gpiochip_irqchip_add() from probe().
>

Ok. I'll do that.

> > -static int gb_gpio_to_irq(struct gpio_chip *chip, unsigned int offset)
> > -{
> > -	struct gb_gpio_controller *ggc = gpio_chip_to_gb_gpio_controller(chip);
> > -
> > -	return irq_find_mapping(ggc->irqdomain, offset);
> > -}
> > -
> >  static int gb_gpio_probe(struct gbphy_device *gbphy_dev,
> >  			 const struct gbphy_device_id *id)
> >  {
> > @@ -693,7 +613,6 @@ static int gb_gpio_probe(struct gbphy_device *gbphy_dev,
> >  	gpio->get = gb_gpio_get;
> >  	gpio->set = gb_gpio_set;
> >  	gpio->set_config = gb_gpio_set_config;
> > -	gpio->to_irq = gb_gpio_to_irq;
> >  	gpio->base = -1;		/* Allocate base dynamically */
> >  	gpio->ngpio = ggc->line_max + 1;
> >  	gpio->can_sleep = true;
> > @@ -702,24 +621,23 @@ static int gb_gpio_probe(struct gbphy_device *gbphy_dev,
> >  	if (ret)
> >  		goto exit_line_free;
> >  
> > -	ret = gb_gpio_irqchip_add(gpio, irqc, 0,
> > -				   handle_level_irq, IRQ_TYPE_NONE);
> > +	ret = gpiochip_add(gpio);
> >  	if (ret) {
> > -		dev_err(&gbphy_dev->dev, "failed to add irq chip: %d\n", ret);
> > +		dev_err(&gbphy_dev->dev, "failed to add gpio chip: %d\n", ret);
> >  		goto exit_line_free;
> >  	}
> >  
> > -	ret = gpiochip_add(gpio);
> > +	ret = gb_gpio_irqchip_add(gpio, irqc, 0,
> > +				  handle_level_irq, IRQ_TYPE_NONE);
> 
> As you may have noted, we had the registration order reversed here to
> handle a (theoretical) race, which may or may not only have been an
> issue for the old 3.10 vendor kernel this was developed against. I've
> forgotten the details, but see:
> 
> 	88f6ba61f25b ("greybus: gpio: create irqdomain before registering gpio controller")
> 
> It's possibly an issue also for mainline, but this is indeed the
> registration order all other drivers use (even if they tend not to be
> hotpluggable).
> 

Yes, I noted that while taking reference of some of the existing
drivers. Thanks for the explanation though.

> >  	if (ret) {
> > -		dev_err(&gbphy_dev->dev, "failed to add gpio chip: %d\n", ret);
> > -		goto exit_gpio_irqchip_remove;
> > +		dev_err(&gbphy_dev->dev, "failed to add irq chip: %d\n", ret);
> > +		gpiochip_remove(gpio);
> 
> Please use an error label for this.
> 

Ok. I'll do that.

> > +		goto exit_line_free;
> >  	}
> >  
> >  	gbphy_runtime_put_autosuspend(gbphy_dev);
> >  	return 0;
> >  
> > -exit_gpio_irqchip_remove:
> > -	gb_gpio_irqchip_remove(ggc);
> >  exit_line_free:
> >  	kfree(ggc->lines);
> >  exit_connection_disable:
> 
> Thanks,
> Johan

Thanks a lot for the review comments.

Regards,
Nishad