demuxing irqs

[Jon Smirl]

On Tue, Sep 16, 2008 at 5:42 PM, Anton Vorontsov
<avorontsov at ru.mvista.com> wrote:
> On Tue, Sep 16, 2008 at 02:32:45PM -0400, Jon Smirl wrote:
>> On Tue, Sep 16, 2008 at 1:49 PM, Scott Wood <scottwood at freescale.com> wrote:
>> > On Tue, Sep 16, 2008 at 10:24:35AM -0400, Jon Smirl wrote:
>> >> On Tue, Sep 16, 2008 at 10:14 AM, Anton Vorontsov
>> >> <avorontsov at ru.mvista.com> wrote:
>> >> > Assume that GPIO 8 does not translate to any IRQ, but IRQ 8 is still
>> >> > valid virq b/c it is mapped for another IRQ controller (particularly
>> >> > lots of kernel code assumes that IRQ 8 is 8259 PIC's CMOS interrupt,
>> >> > the PIC and IRQ8 is widely used on PowerPC).
>> >>
>> >> Set the base in the GPIO struct such that this won't happen.  You can
>> >> set the base greater than MAX_IRQ.
>> >
>> > And then you'll conflict with some other subsystem that decides to engage
>> > in the same shenanigans.
>>
>> That comment was target at GPIO's that don't support interrupts. Give
>> those GPIO numbers greater than MAX_IRQ in case someone tries to use
>> them with the IRQ subsystem. Then they'll get errors.
>
> Or we can do the right thing, without messing all other gpio
> controllers, i.e. implementing MAX_IRQ hacks. Right?
>
> I still don't see any problems with .to_irq callback, can you
> point out any?


You have to map between GPIO and IRQ inside the interrupt handlers so
it has to be reasonably fast. This gets done on every shared interrupt
so you will end up building mapping tables. Also, gpio_to_irq()
doesn't take the gpio chip struct as a parameter.

Why does this mess with all of ther GPIO controllers? If they generate
interrupts they obviously have to coordinate with the VIRQ system.
This may be an issue with the way gpio lib is designed, the API for
the library assumes all gpios in the system are assigned unique
identifiers.

Is there any other problem with 1:1 other than it doesn't return an
error if gpio_to_irq() is called with a gpio number that doesn't
support irqs? You could always implement gpio_to_irq() like this:

if (gpio < MAX_HW_IRQ)
   return -ENOSYSl
return gpio;

Sure your proposal works too, it's just more complicated. 1:1 mapping
is working for ARM, why does PowerPC need to be different? I initially
started coding it the way you propose but then I stumbled across the
ARM solution and it was way simpler.


-- 
Jon Smirl
jonsmirl at gmail.com