Hi Jon,

On Tue, Feb 16, 2016 at 10:28 AM, Jon Hunter [off-list ref] wrote:

On 09/02/16 15:19, Geert Uytterhoeven wrote:

quoted

The R-Car GPIO driver handles Runtime PM for requested GPIOs only.

When using a GPIO purely as an interrupt source, no Runtime PM handling
is done, and the GPIO module's clock may not be enabled.

To fix this:
  - Add .irq_request_resources() and .irq_release_resources() callbacks
    to handle Runtime PM when an interrupt is requested,

I know that when I was looking at runtime-pm support for IRQ chips
(which I have been meaning to get back too), the problem with
irq_request_resources() is that it is called from the context of a
spinlock (in __setup_irq()). You mentioned that you have not seen any
reports of might_sleep_if(), but have you ensured that it is actually
runtime resuming in your testing and you are not getting lucky?

It must be runtime-resuming, because without the call to pm_runtime_get_sync()
interrupts don't work.

An alternative for you might be to use the
irq_bus_lock/irq_bus_sync_unlock callbacks. See what Grygorii
implemented for OMAP in commit aca82d1cbb49 ("gpio: omap: move pm
runtime in irq_chip.irq_bus_lock/sync_unlock").

Thanks, that indeed looks interesting...

As I mentioned I do plan to get back to the series for adding runtime-pm
support for IRQ chips, in the next week or two.

Looking forward to it!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds