On 18/01/2019 12:42, Brian Masney wrote:
> Hi Marc,
>
> On Thu, Jan 17, 2019 at 11:32:01AM +0000, Marc Zyngier wrote:
>>> static int pmic_gpio_to_irq(struct gpio_chip *chip, unsigned pin)
>>> {
>>> struct pmic_gpio_state *state = gpiochip_get_data(chip);
>>> - struct pmic_gpio_pad *pad;
>>> + struct irq_fwspec fwspec;
>>>
>>> - pad = state->ctrl->desc->pins[pin].drv_data;
>>> + fwspec.fwnode = state->fwnode;
>>> + fwspec.param_count = 2;
>>> + fwspec.param[0] = pin + PMIC_GPIO_PHYSICAL_OFFSET;
>>> + fwspec.param[1] = IRQ_TYPE_NONE;
>>
>> In my experience, IRQ_TYPE_NONE is rarely a good thing, unless you
>> expect the trigger information to be found by some other mean. I guess
>> that's one of the reasons why everything falls back to level in the SPMI
>> driver...
>
> I'm not sure how to determine what trigger to put here. I thought that
> it would be up to the caller of request_any_context_irq() to explicitly
> set the expected trigger type when a GPIO is used, which will overwrite
> IRQ_TYPE_NONE with the proper trigger type.
The main issue is that IRQ_TYPE_NONE is a bit loosely defined, and
mostly means "keep whatever was there before", which is a bit like
rolling a dice each time you allocate an interrupt.
>
> For example, I've tested the hierarchical IRQ domains with gpio-keys and
> when the gpio property is used, devm_request_any_context_irq() is called
> with the flags IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING. This calls
> __setup_irq(), which will call irq_set_type() and overwrite the trigger
> type.
>
> irq_set_type() is only called when the IRQ is not shared, so I'm not
> sure if this would work as expected with a shared IRQ.
I'd suggest you force the type to a "safe" value such as rising edge,
and let the irq_set_type() call do the right thing, assuming you've
plugged the issue in the core SPMI driver.
Thanks,
M.
--
Jazz is not dead. It just smells funny...
