Re: [PATCH 06/12] gpio: Add Aspeed driver

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 2016-08-11 at 11:20 +0200, Linus Walleij wrote:
> On Wed, Jul 20, 2016 at 7:58 AM, Andrew Jeffery <andrew@xxxxxxxx>
> wrote:
> 
> > 
> > diff --git a/arch/arm/mach-aspeed/Kconfig b/arch/arm/mach-
> > aspeed/Kconfig
> > index 25a0ae01429e..a52de9d3adfb 100644
> > --- a/arch/arm/mach-aspeed/Kconfig
> > +++ b/arch/arm/mach-aspeed/Kconfig
> > @@ -6,6 +6,10 @@ menuconfig ARCH_ASPEED
> >         select ASPEED_WATCHDOG
> >         select MOXART_TIMER
> >         select PINCTRL
> > +       select GPIOLIB
> > +       select GPIO_ASPEED
> Again this needs to be a separate patch to ARM SoC.
> 
> > 
> > +       select GPIO_SYSFS
> NAK over my dead body. I strongly discourage the use of the
> GPIO sysfs, use the new character device, see
> toos/gpio/* for examples.
> 
> > 
> > +config GPIO_ASPEED
> > +       bool "Aspeed GPIO support"
> > +       depends on (ARCH_ASPEED || COMPILE_TEST) && OF
> > +       select GENERIC_IRQ_CHIP
> Why are you using GENERIC_IRQ_CHIP but not
> GPIOLIB_IRQCHIP? Well I guess I may find out by
> reading the code...
> 
> > 
> > +       help
> > +         Say Y here to support Aspeed AST2400 and AST2500 GPIO
> > controllers.
> > +
> >  config GPIO_BCM_KONA
> >         bool "Broadcom Kona GPIO"
> >         depends on OF_GPIO && (ARCH_BCM_MOBILE || COMPILE_TEST)
> > @@ -1072,7 +1079,6 @@ config GPIO_SODAVILLE
> >         select GENERIC_IRQ_CHIP
> >         help
> >           Say Y here to support Intel Sodaville GPIO.
> > -
> >  endmenu
> Drop this unrelated whitespace change.
> 
> > 
> > +#define GPIO_BANK(x)   ((x) >> 5)
> > +#define GPIO_OFFSET(x) ((x) & 0x1f)
> > +#define GPIO_BIT(x)    BIT(GPIO_OFFSET(x))
> Clever, maybe needs some comments on how they work?
> Or is it obvious from context?
> 
> > 
> > +#define GPIO_DATA      0x00
> > +#define GPIO_DIR       0x04
> > +
> > +#define GPIO_IRQ_ENABLE        0x00
> > +#define GPIO_IRQ_TYPE0 0x04
> > +#define GPIO_IRQ_TYPE1 0x08
> > +#define GPIO_IRQ_TYPE2 0x0c
> > +#define GPIO_IRQ_STATUS        0x10
> > 
> > +static inline struct aspeed_gpio *to_aspeed_gpio(struct gpio_chip
> > *chip)
> > +{
> > +       return container_of(chip, struct aspeed_gpio, chip);
> > +}
> NAK rewrite your code to use devm_gpiochip_add_data() and
> then use gpiochip_get_data() inline in every function that
> needs to get the state container from the gpiochip. Read
> the code in the upstream kernel for *any* driver because I
> think I changed this virtually everywhere.
> 
> (...)
> > 
> > +static void __aspeed_gpio_irq_set_mask(struct irq_data *d, bool
> > set)
> Why __underscoring. It is just confusing, drop the underscores.
> The function does set the mask.
> 
> > 
> > +{
> > +       const struct aspeed_gpio_bank *bank;
> > +       struct aspeed_gpio *gpio;
> > +       unsigned long flags;
> > +       u32 reg, bit;
> > +       void *addr;
> > +       int rc;
> > +
> > +       rc = irqd_to_aspeed_gpio_data(d, &gpio, &bank, &bit);
> > +       if (rc)
> > +               return;
> > +
> > +       addr = bank_irq_reg(gpio, bank, GPIO_IRQ_ENABLE);
> > +
> > +       spin_lock_irqsave(&gpio->lock, flags);
> > +
> > +       reg = ioread32(addr);
> > +       if (set)
> > +               reg |= bit;
> > +       else
> > +               reg &= bit;
> > +       iowrite32(reg, addr);
> Hm, if this was done with regmap it would be regmap_update_bits()
> simply ... maybe you should just throw a 32bit MMIO regmap over
> the registers? (Not required, just an idea to simplify stuff...)
> 
> > 
> > +static int aspeed_gpio_set_type(struct irq_data *d, unsigned int
> > type)
> > +{
> > +       u32 type0, type1, type2, bit, reg;
> > +       const struct aspeed_gpio_bank *bank;
> > +       irq_flow_handler_t handler;
> > +       struct aspeed_gpio *gpio;
> > +       unsigned long flags;
> > +       void *addr;
> > +       int rc;
> > +
> > +       rc = irqd_to_aspeed_gpio_data(d, &gpio, &bank, &bit);
> > +       if (rc)
> > +               return -EINVAL;
> > +
> > +       type0 = type1 = type2 = 0;
> Assign zero when declaring instead:
> 
> u32 type0 = 0;
> u32 type1 = 0;
> ...
> 
> > 
> > +
> > +       switch (type & IRQ_TYPE_SENSE_MASK) {
> > +       case IRQ_TYPE_EDGE_BOTH:
> > +               type2 |= bit;
> > +       case IRQ_TYPE_EDGE_RISING:
> > +               type0 |= bit;
> > +       case IRQ_TYPE_EDGE_FALLING:
> > +               handler = handle_edge_irq;
> > +               break;
> > +       case IRQ_TYPE_LEVEL_HIGH:
> > +               type0 |= bit;
> > +       case IRQ_TYPE_LEVEL_LOW:
> > +               type1 |= bit;
> > +               handler = handle_level_irq;
> > +               break;
> > +       default:
> > +               return -EINVAL;
> > +       }
> Very nice, this looks exactly as it should, handling the different
> IRQs very nicely.
> 
> > 
> > +       spin_lock_irqsave(&gpio->lock, flags);
> > +
> > +       addr = bank_irq_reg(gpio, bank, GPIO_IRQ_TYPE0);
> > +       reg = ioread32(addr);
> > +       reg = (reg & ~bit) | type0;
> > +       iowrite32(reg, addr);
> > +
> > +       addr = bank_irq_reg(gpio, bank, GPIO_IRQ_TYPE1);
> > +       reg = ioread32(addr);
> > +       reg = (reg & ~bit) | type1;
> > +       iowrite32(reg, addr);
> > +
> > +       addr = bank_irq_reg(gpio, bank, GPIO_IRQ_TYPE2);
> > +       reg = ioread32(addr);
> > +       reg = (reg & ~bit) | type2;
> > +       iowrite32(reg, addr);
> > +
> > +       spin_unlock_irqrestore(&gpio->lock, flags);
> > +
> > +       irq_set_handler_locked(d, handler);
> > +
> > +       return 0;
> > +}
> Overall very nice .set_type().
> 
> > 
> > +static void aspeed_gpio_irq_handler(struct irq_desc *desc)
> > +{
> > +       struct aspeed_gpio *gpio = irq_desc_get_handler_data(desc);
> > +       struct irq_chip *chip = irq_desc_get_chip(desc);
> > +       unsigned int i, p, girq;
> > +       unsigned long reg;
> > +
> > +       chained_irq_enter(chip, desc);
> > +
> > +       for (i = 0; i < ARRAY_SIZE(aspeed_gpio_banks); i++) {
> > +               const struct aspeed_gpio_bank *bank =
> > &aspeed_gpio_banks[i];
> > +
> > +               reg = ioread32(bank_irq_reg(gpio, bank,
> > GPIO_IRQ_STATUS));
> > +
> > +               for_each_set_bit(p, &reg, 32) {
> > +                       girq = irq_find_mapping(gpio->irq_domain, i
> > * 32 + p);
> > +                       generic_handle_irq(girq);
> > +               }
> > +
> > +       }
> > +
> > +       chained_irq_exit(chip, desc);
> > +}
> This looks so generic so I think you should be using
> GPIOLIB_IRQCHIP.
> 
> > 
> > +
> > +static struct irq_chip aspeed_gpio_irqchip = {
> > +       .name           = "aspeed-gpio",
> > +       .irq_ack        = aspeed_gpio_irq_ack,
> > +       .irq_mask       = aspeed_gpio_irq_mask,
> > +       .irq_unmask     = aspeed_gpio_irq_unmask,
> > +       .irq_set_type   = aspeed_gpio_set_type,
> > +};
> There is a missing .request/.release resources marking the lines
> as irq to the GPIO core. But if you switch to using GPIOLIB_IRQCHIP
> the core will handle all that for you.
> 
> > 
> > +static int aspeed_gpio_to_irq(struct gpio_chip *chip, unsigned int
> > offset)
> > +{
> > +       struct aspeed_gpio *gpio = to_aspeed_gpio(chip);
> > +
> > +       return irq_find_mapping(gpio->irq_domain, offset);
> > +}
> > +
> > +static void aspeed_gpio_setup_irqs(struct aspeed_gpio *gpio,
> > +               struct platform_device *pdev)
> > +{
> > +       int i, irq;
> > +
> > +       /* request our upstream IRQ */
> > +       gpio->irq = platform_get_irq(pdev, 0);
> > +       if (gpio->irq < 0)
> > +               return;
> > +
> > +       /* establish our irq domain to provide IRQs for each
> > extended bank */
> > +       gpio->irq_domain = irq_domain_add_linear(pdev->dev.of_node,
> > +                       gpio->chip.ngpio, &irq_domain_simple_ops,
> > NULL);
> > +       if (!gpio->irq_domain)
> > +               return;
> > +
> > +       for (i = 0; i < gpio->chip.ngpio; i++) {
> > +               irq = irq_create_mapping(gpio->irq_domain, i);
> > +               irq_set_chip_data(irq, gpio);
> > +               irq_set_chip_and_handler(irq, &aspeed_gpio_irqchip,
> > +                               handle_simple_irq);
> > +               irq_set_probe(irq);
> > +       }
> > +
> > +       irq_set_chained_handler_and_data(gpio->irq,
> > +                       aspeed_gpio_irq_handler, gpio);
> > +}
> Also all this goes away with GPILIB_IRQCHIP, so use it.
> 
> See e.g. drivers/gpio/gpio-pl061.c for an example of a similar
> driver doing this.
> 
> > 
> > +static int __init aspeed_gpio_probe(struct platform_device *pdev)
> > +{
> > +       struct aspeed_gpio *gpio;
> > +       struct resource *res;
> > +       int rc;
> > +
> > +       gpio = devm_kzalloc(&pdev->dev, sizeof(*gpio), GFP_KERNEL);
> > +       if (!gpio)
> > +               return -ENOMEM;
> > +
> > +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > +       if (!res)
> > +               return -ENXIO;
> > +
> > +       gpio->base = devm_ioremap_resource(&pdev->dev, res);
> > +       if (!gpio->base)
> > +               return -ENOMEM;
> > +
> > +       spin_lock_init(&gpio->lock);
> > +
> > +       gpio->chip.ngpio = ARRAY_SIZE(aspeed_gpio_banks) * 32;
> > +
> > +       gpio->chip.parent = &pdev->dev;
> > +       gpio->chip.direction_input = aspeed_gpio_dir_in;
> > +       gpio->chip.direction_output = aspeed_gpio_dir_out;
> Please add gpio->chip.get_direction() to complete the picture.
> 
> > 
> > +       gpio->chip.request = aspeed_gpio_request;
> > +       gpio->chip.free = aspeed_gpio_free;
> > +       gpio->chip.get = aspeed_gpio_get;
> > +       gpio->chip.set = aspeed_gpio_set;
> > +       gpio->chip.to_irq = aspeed_gpio_to_irq;
> .to_irq goes away with GPILIB_IRQCHIP.
> 
> > 
> > +       gpio->chip.label = dev_name(&pdev->dev);
> > +       gpio->chip.base = -1;
> > +
> > +       platform_set_drvdata(pdev, gpio);
> > +
> > +       rc = gpiochip_add(&gpio->chip);
> > +       if (rc < 0)
> > +               return rc;
> Use devm_gpiochip_add_data(), then gpiochip_irqchip_add().
> 
> > 
> > +static int aspeed_gpio_remove(struct platform_device *pdev)
> > +{
> > +       struct aspeed_gpio *gpio = platform_get_drvdata(pdev);
> > +
> > +       gpiochip_remove(&gpio->chip);
> > +       return 0;
> > +}
> And then this is not even needed. (devm*)
> 
> Yours,
> Linus Walleij

Thanks, I will clean up the issues. I brought the patch into the series
because I figured that's what made sense. I should have looked at it a
bit closer.

Cheers,

Andrew

Attachment: signature.asc
Description: This is a digitally signed message part


[Index of Archives]     [Linux SPI]     [Linux Kernel]     [Linux ARM (vger)]     [Linux ARM MSM]     [Linux Omap]     [Linux Arm]     [Linux Tegra]     [Fedora ARM]     [Linux for Samsung SOC]     [eCos]     [Linux Fastboot]     [Gcc Help]     [Git]     [DCCP]     [IETF Announce]     [Security]     [Linux MIPS]     [Yosemite Campsites]

  Powered by Linux