[3/4] gpio: vf610: add gpiolib/IRQ chip driver for Vybird

Add a gpiolib and IRQ chip driver for Vybrid ARM SoC using the
Vybrid's GPIO and PORT module. The driver is instanced once per
each GPIO/PORT module pair and handles 32 GPIO's.

Signed-off-by: Stefan Agner <stefan@agner.ch>
---
 drivers/gpio/Kconfig      |   7 ++
 drivers/gpio/Makefile     |   1 +
 drivers/gpio/gpio-vf610.c | 285 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 293 insertions(+)
 create mode 100644 drivers/gpio/gpio-vf610.c

On Sat, Sep 6, 2014 at 6:25 PM, Stefan Agner <stefan@agner.ch> wrote:

> Add a gpiolib and IRQ chip driver for Vybrid ARM SoC using the
> Vybrid's GPIO and PORT module. The driver is instanced once per
> each GPIO/PORT module pair and handles 32 GPIO's.
>
> Signed-off-by: Stefan Agner <stefan@agner.ch>
(...)
>  obj-$(CONFIG_GPIO_UCB1400)     += gpio-ucb1400.o
> +obj-$(CONFIG_GPIO_VF610)       += gpio-vf610.o

Some like to keep GPIOs tightly associated with a pin controller
in a file next to the pin controller.

I.e. in drivers/pinctrl/freescale/gpio-vf610.c

But this works too. Any preference?

> +#define GPIO_PER_PORT          32

Very generic define. VF610_GPIOS_PER_PORT?

> +struct vf610_gpio_port {
> +       struct gpio_chip gc;
> +       void __iomem *base;
> +       void __iomem *gpio_base;
> +       u8 irqc[GPIO_PER_PORT];
> +       int irq;

irq? Why do you need to keep this around?

> +static const struct of_device_id vf610_gpio_dt_ids[] = {
> +       { .compatible = "fsl,vf610-gpio" },
> +       { /* sentinel */ }
> +};
> +
> +static inline void vf610_gpio_writel(u32 val, void __iomem *reg)
> +{
> +       __raw_writel(val, reg);

Use writel_relaxed() instead unless you can explain why you want this.

(Same for all occurences.)

> +static int vf610_gpio_get(struct gpio_chip *gc, unsigned int gpio)
> +{
> +       struct vf610_gpio_port *port =
> +               container_of(gc, struct vf610_gpio_port, gc);
> +
> +       return !!(vf610_gpio_readl(port->gpio_base + GPIO_PDIR) & 1 << gpio);

#include <linux/bitops.h>

... & BIT(gpio)

> +static void vf610_gpio_set(struct gpio_chip *gc, unsigned int gpio, int val)
> +{
> +       struct vf610_gpio_port *port =
> +               container_of(gc, struct vf610_gpio_port, gc);
> +       unsigned long mask = 1 << gpio;

= BIT(gpio);

> +static void vf610_gpio_irq_handler(u32 irq, struct irq_desc *desc)
> +{
> +       struct vf610_gpio_port *port = irq_get_handler_data(irq);
> +       struct irq_chip *chip = irq_desc_get_chip(desc);
> +       int pin;
> +       unsigned long irq_isfr;
> +
> +       chained_irq_enter(chip, desc);
> +
> +       irq_isfr = vf610_gpio_readl(port->base + PORT_ISFR);
> +
> +       for_each_set_bit(pin, &irq_isfr, GPIO_PER_PORT) {
> +               vf610_gpio_writel(1 << pin, port->base + PORT_ISFR);

BIT(pin)

(etc)

> +       port->irq = irq_of_parse_and_map(pdev->dev.of_node, 0);
> +       if (port->irq == NO_IRQ)
> +               return -ENODEV;

Can't you just use a local int irq; variable for this?

> +static int __init gpio_vf610_init(void)
> +{
> +       return platform_driver_register(&vf610_gpio_driver);
> +}
> +postcore_initcall(gpio_vf610_init);

postcore again. I don't like this, can you get rid of it?

Overall the driver looks very nice except for these nitty gritty details.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Hi Linus,

Thanks for your review!

Am 2014-09-23 11:58, schrieb Linus Walleij:
> On Sat, Sep 6, 2014 at 6:25 PM, Stefan Agner <stefan@agner.ch> wrote:
> 
>> Add a gpiolib and IRQ chip driver for Vybrid ARM SoC using the
>> Vybrid's GPIO and PORT module. The driver is instanced once per
>> each GPIO/PORT module pair and handles 32 GPIO's.
>>
>> Signed-off-by: Stefan Agner <stefan@agner.ch>
> (...)
>>  obj-$(CONFIG_GPIO_UCB1400)     += gpio-ucb1400.o
>> +obj-$(CONFIG_GPIO_VF610)       += gpio-vf610.o
> 
> Some like to keep GPIOs tightly associated with a pin controller
> in a file next to the pin controller.
> 
> I.e. in drivers/pinctrl/freescale/gpio-vf610.c
> 
> But this works too. Any preference?

The other Freescale GPIO drivers (e.g. gpio-mxs.c/gpio-mxc.c) are
located under drivers/gpio/ hence I would prefer to leave them there,
even we use pinctrl here. Unless someone at Freescale has another
opinion on this?


> 
>> +#define GPIO_PER_PORT          32
> 
> Very generic define. VF610_GPIOS_PER_PORT?

Agreed

> 
>> +struct vf610_gpio_port {
>> +       struct gpio_chip gc;
>> +       void __iomem *base;
>> +       void __iomem *gpio_base;
>> +       u8 irqc[GPIO_PER_PORT];
>> +       int irq;
> 
> irq? Why do you need to keep this around?
> 
>> +static const struct of_device_id vf610_gpio_dt_ids[] = {
>> +       { .compatible = "fsl,vf610-gpio" },
>> +       { /* sentinel */ }
>> +};
>> +
>> +static inline void vf610_gpio_writel(u32 val, void __iomem *reg)
>> +{
>> +       __raw_writel(val, reg);
> 
> Use writel_relaxed() instead unless you can explain why you want this.
> 
> (Same for all occurences.)

Agreed, I have don't know why I used the __raw variant here. I think its
because copied this two stubs from gpio-tegra.c.

> 
>> +static int vf610_gpio_get(struct gpio_chip *gc, unsigned int gpio)
>> +{
>> +       struct vf610_gpio_port *port =
>> +               container_of(gc, struct vf610_gpio_port, gc);
>> +
>> +       return !!(vf610_gpio_readl(port->gpio_base + GPIO_PDIR) & 1 << gpio);
> 
> #include <linux/bitops.h>
> 
> ... & BIT(gpio)
> 
>> +static void vf610_gpio_set(struct gpio_chip *gc, unsigned int gpio, int val)
>> +{
>> +       struct vf610_gpio_port *port =
>> +               container_of(gc, struct vf610_gpio_port, gc);
>> +       unsigned long mask = 1 << gpio;
> 
> = BIT(gpio);
> 
>> +static void vf610_gpio_irq_handler(u32 irq, struct irq_desc *desc)
>> +{
>> +       struct vf610_gpio_port *port = irq_get_handler_data(irq);
>> +       struct irq_chip *chip = irq_desc_get_chip(desc);
>> +       int pin;
>> +       unsigned long irq_isfr;
>> +
>> +       chained_irq_enter(chip, desc);
>> +
>> +       irq_isfr = vf610_gpio_readl(port->base + PORT_ISFR);
>> +
>> +       for_each_set_bit(pin, &irq_isfr, GPIO_PER_PORT) {
>> +               vf610_gpio_writel(1 << pin, port->base + PORT_ISFR);
> 
> BIT(pin)
> 
> (etc)

Ok, will replace those bit operations.

> 
>> +       port->irq = irq_of_parse_and_map(pdev->dev.of_node, 0);
>> +       if (port->irq == NO_IRQ)
>> +               return -ENODEV;
> 
> Can't you just use a local int irq; variable for this?
> 
>> +static int __init gpio_vf610_init(void)
>> +{
>> +       return platform_driver_register(&vf610_gpio_driver);
>> +}
>> +postcore_initcall(gpio_vf610_init);
> 
> postcore again. I don't like this, can you get rid of it?

I guess we could load this driver easily a bit later. IMHO, since lots
of other driver use GPIO's, we should it load before all the drivers
gets loaded (before device_initcall).

Most GPIO driver do this, some statistic again:
$ grep -h -o ".*_initcall" drivers/gpio/*.c | sort | uniq -c | sort -n
-r
     33 subsys_initcall
     14 postcore_initcall
      2 device_initcall
      2 arch_initcall
      1 late_initcall
      1 core_initcall

My proposal:
Use subsys_initcall (which is called after arch_initcall) in this GPIO
driver and leave the pinctrl driver as arch_initcall. This way we are
absolutely sure that the GPIO driver gets loaded after the pinctrl and
also leave the pinctrl at its current value.

--
Stefan


> 
> Overall the driver looks very nice except for these nitty gritty details.
> 
> Yours,
> Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Tue, Sep 23, 2014 at 1:51 PM, Stefan Agner <stefan@agner.ch> wrote:
> [Me]
>> postcore again. I don't like this, can you get rid of it?
>
> I guess we could load this driver easily a bit later. IMHO, since lots
> of other driver use GPIO's, we should it load before all the drivers
> gets loaded (before device_initcall).

Nope. We use deferred probing to control that today. Ideally
all drivers should be device_initcall() and deferred probe be used
to order things, not by playing around with initcalls.

> Most GPIO driver do this, some statistic again:
> $ grep -h -o ".*_initcall" drivers/gpio/*.c | sort | uniq -c | sort -n
> -r
>      33 subsys_initcall
>      14 postcore_initcall
>       2 device_initcall
>       2 arch_initcall
>       1 late_initcall
>       1 core_initcall

Yeah old legacy. There are patch attacks to get rid of this.

The reason we can't just change them is because sometimes
dependent drivers do not handle the errorpath very well can can't
defer cleanly.

With a new driver I expect deferred probe to be used.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Am 2014-09-24 13:10, schrieb Linus Walleij:
> On Tue, Sep 23, 2014 at 1:51 PM, Stefan Agner <stefan@agner.ch> wrote:
>> [Me]
>>> postcore again. I don't like this, can you get rid of it?
>>
>> I guess we could load this driver easily a bit later. IMHO, since lots
>> of other driver use GPIO's, we should it load before all the drivers
>> gets loaded (before device_initcall).
> 
> Nope. We use deferred probing to control that today. Ideally
> all drivers should be device_initcall() and deferred probe be used
> to order things, not by playing around with initcalls.
> 

You can not "nope" my humble opinion, this is not possible, its write
protected! :-)

I fully understand the deferred probe approach, and I also think its
really a good approach for almost all drivers. The system by itself
makes sure the drivers are loaded in correct order. But if all drivers
use pinctrl, and the pinctrl happens to be the last initialized driver,
I first get 30 messages with probe deferred before the pinctrl driver
finally gets initialized. IMHO, this is a waste of resources... Giving
the system some hint what we need early (e.g. pinctrl or even GPIO
drivers) is just sensible. But maybe I miss something here...

Fact is, currently, without touching GPIO infrastructure code, the GPIO
driver can not be deferred when the pinctrl driver is missing. The
of_gpiochip_add_pin_range function does not handle the missing pinctrl
driver. Hence as of now, the pinctrl still needs an earlier initcall.
But anyway, currently the pinctrl driver is already using arch_initcall,
this patch set is no longer touching that.

>> Most GPIO driver do this, some statistic again:
>> $ grep -h -o ".*_initcall" drivers/gpio/*.c | sort | uniq -c | sort -n
>> -r
>>      33 subsys_initcall
>>      14 postcore_initcall
>>       2 device_initcall
>>       2 arch_initcall
>>       1 late_initcall
>>       1 core_initcall
> 
> Yeah old legacy. There are patch attacks to get rid of this.
> 
> The reason we can't just change them is because sometimes
> dependent drivers do not handle the errorpath very well can can't
> defer cleanly.
> 
> With a new driver I expect deferred probe to be used.
> 

Actually, the esdhc driver gets the cd-gpio directly and does not handle
EPROBE_DEFER properly, hence Vybrid would be affected too. But I
understand that we need to start somewhere. I will change the GPIO
driver to device_initcall and going to fix esdhc driver.

--
Stefan


--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html