[RFC PATCH] gpio/omap: Fix IRQ handling for SPARSE_IRQ

[DebBarma, Tarun Kanti]

On Fri, Feb 24, 2012 at 7:02 PM, Cousson, Benoit <b-cousson at ti.com> wrote:
> On 2/24/2012 2:24 PM, DebBarma, Tarun Kanti wrote:
>>
>> Hi Benoit,
>>
>> On Fri, Feb 24, 2012 at 4:07 PM, DebBarma, Tarun Kanti
>> <tarun.kanti at ti.com>  wrote:
>>>
>>> On Fri, Feb 24, 2012 at 3:41 PM, Cousson, Benoit<b-cousson at ti.com>
>>>  wrote:
>>>>
>>>> + Tarun
>>>>
>>>>
>>>> On 2/24/2012 12:08 AM, Tony Lindgren wrote:
>>>>>
>>>>>
>>>>> * Cousson, Benoit<b-cousson at ti.com>    [120223 14:14]:
>>>>>>
>>>>>>
>>>>>> The GPIO driver is still relying on internal OMAP IRQ defines that
>>>>>> are not relevant anymore if OMAP is built with SPARSE_IRQ.
>>>>>
>>>>>
>>>>>
>>>>> Great :)
>>>>>
>>>>>> Please note that this patch is still RFC, because I do not know
>>>>>> how to fix properly the ugly cpu_class_is_omap1 and the dependency
>>>>>> with IH_MPUIO_BASE to detect a MPUIO.
>>>>>
>>>>>
>>>>>
>>>>> Sounds like gpio_to_irq() needs to be set in the
>>>>> arch/arm/*omap*/gpio*.c then.
>>>>
>>>>
>>>>
>>>> In fact, after a second thought, that might even work for OMAP1 because
>>>> I'm
>>>> using the proper base (IRQ and GPIO) to convert the IRQ number.
>>>>
>>>>
>>>> static int irq_to_gpio(struct gpio_bank *bank, unsigned int gpio_irq)
>>>> {
>>>>
>>>>        return gpio_irq - bank->irq_base + bank->chip.base;
>>>> }
>>>>
>>>> But it might be good to test it on OMAP1 platform.
>>>>
>>>>
>>>> Tarun,
>>>>
>>>> Do you have an OMAP1 board to test that.
>>>
>>> Yes, I will test on OMAP1 board.
>>
>> I have booted the image on OMAP1 with following change.
>> I guess bank->irq_base was a typo?
>
>
> Not at all :-), it was done on purpose to get rid if the static IRQ
> definition.
BTW, it was giving me following compilation error...
drivers/gpio/gpio-omap.c: In function 'irq_to_gpio':
drivers/gpio/gpio-omap.c:90: error: 'struct gpio_bank' has no member
named 'irq_base'
make[2]: *** [drivers/gpio/gpio-omap.o] Error 1
--
Tarun

> I was expecting that kind of issue because OMAP1 is populating
> virtual_irq_start with the IH_MPUIO_BASE, but was expecting the dynamic
> alloc_decs to still provide the good value. The easy fix is to still use
> that for OMAP1 GPIO since nobody will want to spend time fixing OMAP1 boards
> to do that properly :-(
>
> I'll update the patch accordingly.
>
>
>> static int irq_to_gpio(struct gpio_bank *bank, unsigned int gpio_irq)
>> {
>>         //return gpio_irq - bank->irq_base + bank->chip.base;
>>         return gpio_irq - bank->virtual_irq_start + bank->chip.base;
>> }
>> --
>
>
> Thanks for the help,
> Benoit
>
>