Hi Sekhar, Linus,

On 11/06/2013 12:49 PM, Sekhar Nori wrote:

On Wednesday 06 November 2013 03:45 PM, Prabhakar Lad wrote:

quoted

Hi Linus,

On Wed, Nov 6, 2013 at 3:26 PM, Linus Walleij [off-list ref] wrote:

quoted

On Wed, Nov 6, 2013 at 10:33 AM, Prabhakar Lad
[off-list ref] wrote:

quoted

On Tue, Nov 5, 2013 at 6:09 PM, Linus Walleij [off-list ref] wrote:

quoted

On Sat, Nov 2, 2013 at 4:39 PM, Lad, Prabhakar
[off-list ref] wrote:

quoted

From: "Lad, Prabhakar" <prabhakar.csengg@gmail.com>

This patch fixes the check for the offset in
gpio_to_irq_unbanked() function.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>

Is this a regression that should go in right now?

Yes it needs too.

But on top of *what* exactly?

This does not apply to my gpio tree devel branch and
not even on the mainline kernel.

Looks like this needs to go via ARM tree as the earlier
patches have  gone via ARM tree itself [1].
If you can ACK it Sekhar can get it in via ARM tree.

The dependent patches are all in linux-next through ARM SoC queued for
v3.13 merge. This fix can either be sent late in merge cycle once Linus
has pulled ARM SoC or after v3.13-rc1 comes out.

Pls, wait a bit - this fix is incomplete :( The below changes have to be 
done too:

diff --git a/drivers/gpio/gpio-davinci.c b/drivers/gpio/gpio-davinci.c
index 6c90cfb..05dbadb 100644
--- a/drivers/gpio/gpio-davinci.c
+++ b/drivers/gpio/gpio-davinci.c

@@ -328,7 +328,7 @@ static int gpio_to_irq_unbanked(struct gpio_chip 

*chip, unsigned offset)
          * can provide direct-mapped IRQs to AINTC (up to 32 GPIOs).
          */
         if (offset < d->gpio_unbanked)
-               return d->gpio_irq + offset;
+               return d->irq_base + offset;
         else
                 return -ENODEV;
  }

@@ -341,7 +341,7 @@ static int gpio_irq_type_unbanked(struct irq_data 

*data, unsigned trigger)

         d = (struct davinci_gpio_controller *)data->handler_data;
         g = (struct davinci_gpio_regs __iomem *)d->regs;
-       mask = __gpio_mask(data->irq - d->gpio_irq);
+       mask = __gpio_mask(data->irq - d->irq_base);

         if (trigger & ~(IRQ_TYPE_EDGE_FALLING | IRQ_TYPE_EDGE_RISING))
                 return -EINVAL;

@@ -419,6 +419,8 @@ static int davinci_gpio_irq_setup(struct 

platform_device *pdev)

                 /* pass "bank 0" GPIO IRQs to AINTC */
                 chips[0].chip.to_irq = gpio_to_irq_unbanked;
+               chips[0].irq_base = bank_irq;
+               chips[0].gpio_unbanked = pdata->gpio_unbanked;
                 binten = BIT(0);

                 /* AINTC handles mask/unmask; GPIO handles triggering */

diff --git a/include/linux/platform_data/gpio-davinci.h 

b/include/linux/platform_data/gpio-davinci.h
index 6efd202..711a002 100644

--- a/include/linux/platform_data/gpio-davinci.h
+++ b/include/linux/platform_data/gpio-davinci.h

@@ -42,7 +42,6 @@ struct davinci_gpio_controller {
         void __iomem            *clr_data;
         void __iomem            *in_data;
         int                     gpio_unbanked;
-       unsigned                gpio_irq;
  };

==================================================================
Also all places where davinci_gpio_register() is called need to be 
updated to use instead of "sizeof". Like:

diff --git a/arch/arm/mach-davinci/dm355.c b/arch/arm/mach-davinci/dm355.c
index ef9ff1f..10334a4 100644
--- a/arch/arm/mach-davinci/dm355.c
+++ b/arch/arm/mach-davinci/dm355.c

@@ -906,8 +906,8 @@ static struct davinci_gpio_platform_data 

dm355_gpio_platform_data = {
  int __init dm355_gpio_register(void)
  {
         return davinci_gpio_register(dm355_gpio_resources,
-                                    sizeof(dm355_gpio_resources),
-                                    &dm355_gpio_platform_data);
+                                     ARRAY_SIZE(dm355_gpio_resources),
+                                     &dm355_gpio_platform_data);
  }


Regards,
- grygorii