[PATCH] ARM: tegra: Fix <mach/gpio.h> to have necessary #includes

[PATCH] ARM: tegra: Fix <mach/gpio.h> to have necessary #includes

[Stephen Warren]

Without this, EINVAL isn't defined, and type bool isn't declared.
This leads to compilation failures.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
---
 arch/arm/mach-tegra/include/mach/gpio.h |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/arch/arm/mach-tegra/include/mach/gpio.h b/arch/arm/mach-tegra/include/mach/gpio.h
index e353805..96af4ff 100644
--- a/arch/arm/mach-tegra/include/mach/gpio.h
+++ b/arch/arm/mach-tegra/include/mach/gpio.h
@@ -20,7 +20,8 @@
 #ifndef __MACH_TEGRA_GPIO_H
 #define __MACH_TEGRA_GPIO_H
 
-#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/types.h>
 #include <mach/irqs.h>
 
 #define TEGRA_NR_GPIOS		INT_GPIO_NR
-- 
1.7.0.4

Re: [PATCH] ARM: tegra: Fix <mach/gpio.h> to have necessary #includes

[Russell King - ARM Linux]

On Thu, Aug 11, 2011 at 12:05:52PM -0600, Stephen Warren wrote:
> Without this, EINVAL isn't defined, and type bool isn't declared.
> This leads to compilation failures.

An alternative approach would be to convert tegra to be able to use
the gpiolib __gpio_to_irq() for the on-SoC stuff, and eliminate
irq_to_gpio().

Would it be possible to do that instead?

RE: [PATCH] ARM: tegra: Fix <mach/gpio.h> to have necessary #includes

[Stephen Warren]

Russell King wrote at Thursday, August 11, 2011 1:36 PM:
> On Thu, Aug 11, 2011 at 12:05:52PM -0600, Stephen Warren wrote:
> > Without this, EINVAL isn't defined, and type bool isn't declared.
> > This leads to compilation failures.
> 
> An alternative approach would be to convert tegra to be able to use
> the gpiolib __gpio_to_irq() for the on-SoC stuff, and eliminate
> irq_to_gpio().
> 
> Would it be possible to do that instead?

That should work too.

I posted a couple patches last Friday to remove irq_to_gpio:

https://lkml.org/lkml/2011/8/5/343
https://lkml.org/lkml/2011/8/5/342

Later in the thread, you said you would merge them into your branch to
avoid merge conflicts with the other GPIO work you were doing.

For gpio_to_irq, yes, I believe we get rid of it pretty easily; we'd
just need to implement gpio-tegra.c's gpio_chip.to_irq function. Should
I cook up a patch based on top of linux-next plus the two patches above
to do that instead?

Thanks.

-- 
nvpublic

Re: [PATCH] ARM: tegra: Fix <mach/gpio.h> to have necessary #includes

[Russell King - ARM Linux]

On Thu, Aug 11, 2011 at 12:49:47PM -0700, Stephen Warren wrote:
> Russell King wrote at Thursday, August 11, 2011 1:36 PM:
> > On Thu, Aug 11, 2011 at 12:05:52PM -0600, Stephen Warren wrote:
> > > Without this, EINVAL isn't defined, and type bool isn't declared.
> > > This leads to compilation failures.
> > 
> > An alternative approach would be to convert tegra to be able to use
> > the gpiolib __gpio_to_irq() for the on-SoC stuff, and eliminate
> > irq_to_gpio().
> > 
> > Would it be possible to do that instead?
> 
> That should work too.
> 
> I posted a couple patches last Friday to remove irq_to_gpio:
> 
> https://lkml.org/lkml/2011/8/5/343
> https://lkml.org/lkml/2011/8/5/342
> 
> Later in the thread, you said you would merge them into your branch to
> avoid merge conflicts with the other GPIO work you were doing.

I suggested it would be a good idea.  Now that I've published those
changes, any chance of having some updated patches in the patch system
please?

> For gpio_to_irq, yes, I believe we get rid of it pretty easily; we'd
> just need to implement gpio-tegra.c's gpio_chip.to_irq function. Should
> I cook up a patch based on top of linux-next plus the two patches above
> to do that instead?

Yes please, again eventually to the patch system once Grant has acked it
to avoid the otherwise certain conflicts.

Thanks for doing this.