uses of irq_to_gpio() in drivers/staging/iio

uses of irq_to_gpio() in drivers/staging/iio

[Ben Dooks]

Whilst looking in drivers/staging/iio/accel/lis3l02dq_core.c, I
came across the following lines of code

    708          if (spi->irq && gpio_is_valid(irq_to_gpio(spi->irq)) > 0) {
    709                  ret = request_threaded_irq(st->us->irq,
    710                                             &lis3l02dq_th,
    711 
&lis3l02dq_event_handler,
    712                                             IRQF_TRIGGER_RISING,
    713                                             "lis3l02dq",
    714                                             indio_dev);

Which gives rise to the following questions:

1) IRQ0 and GPIO0 are often valid numbers for GPIOs.

2) Not all interrupts are necessarily GPIO interrupts. What happens
    if the device is attached to a hardware interrupt line which is
    not a GPIO?

-- 
Ben Dooks <bjdooks@googlemail.com> http://www.fluff.org/ben/

Re: uses of irq_to_gpio() in drivers/staging/iio

[Jonathan Cameron]

On 08/02/11 17:43, Ben Dooks wrote:
> Whilst looking in drivers/staging/iio/accel/lis3l02dq_core.c, I
> came across the following lines of code
> 
>    708          if (spi->irq && gpio_is_valid(irq_to_gpio(spi->irq)) > 0) {
>    709                  ret = request_threaded_irq(st->us->irq,
>    710                                             &lis3l02dq_th,
>    711 &lis3l02dq_event_handler,
>    712                                             IRQF_TRIGGER_RISING,
>    713                                             "lis3l02dq",
>    714                                             indio_dev);
> 
> Which gives rise to the following questions:
> 
> 1) IRQ0 and GPIO0 are often valid numbers for GPIOs.
Good point - should fix that one.  Any suggestions for a neat way
to do it?
> 
> 2) Not all interrupts are necessarily GPIO interrupts. What happens
>    if the device is attached to a hardware interrupt line which is
>    not a GPIO?
> 
Easy. It doesn't work.  Stupid hardware doesn't necessarily drop
the interrupt line on a read. Hence it sticks high. If one only has
edge triggered interrupt lines that's a pain. Is there an easy way
around this?  When I wrote the relevant code a long time ago I couldn't
get anything else to work.

There may well be a better way but I haven't revisited this recently.