Javier,

On Fri, Aug 8, 2014 at 3:26 PM, Javier Martinez Canillas
[off-list ref] wrote:

quoted

I have some vague recollection that I set interrupts to pin-function
"0" by default for some reason (assuming they would go to 0xf when
interrupts were enabled).  ...but I can't for the life of me remember
if it was a good reason or just seemed like the right thing to do.

It would be great to know if there is a good reason. I see indeed that all
pinctrl setup in the Peach Pit/Pi and Snow DTS that involves interrupts are
using 0x0 as the pin function. Since as Tomasz explained Samsung SoC makes a
difference between GPIO-IRQ and GPIO input I guess that it's better to change
those to 0xf instead. What do you think?

I think it's worth trying out.  If there are no problems with it then
let's do it.

My vague recollection is that I was worried that pinctrl would take
effect right at driver probe time (maybe this used to happen? or maybe
I imagined it?) and that configuring to 0xf at this point in time
would cause a spurious interrupt.  I can't remember ever testing it so
it was probably just something I imagined.  Even if it was configured
as 0xf I'd imagine that the interrupt would be masked anyway so there
should be no spurious interrupt, right?

Regardless of this though I think that both the patch to move the IRQ
pinmux setup from .irq_set_type to the .irq_request_resources and the patch to
to prevent any pinmux reconfiguration are good improvements to avoid future
issues like the one we found here.

OK.  I'll let you, Tomasz, and Linus figure out what's best here since
I haven't done extensive thinking on it.  ;)

-Doug