On Mon, May 25, 2020 at 02:35:51PM +0300, Andy Shevchenko wrote: > On Mon, May 25, 2020 at 02:05:56PM +0300, Mika Westerberg wrote: > > On Mon, May 25, 2020 at 01:13:35PM +0300, Andy Shevchenko wrote: > > > Due to parsing of ACPI tables. I don't want to copy'n'paste 25% of > > > gpiolib-acpi.c in here. I think provided solution is cleaner and (more) > > > flexible in terms of maintenance. > > > > Hmm, you seem to pass a hard-coded pin number (1) to the core that then > > passes it back to the driver. Why you can't simple use that number here > > directly? You don't need to parse anything. What I'm missing? :-) > > Okay, so, AFAIU you are proposing something like this: > > 1) find a GPIO controller by the ACPI path (somehow, I guess by finding a > handle followed by physical device behind it); 2) somehow to request a > pin from that device by number; > 3) convert to IRQ and use. > > Is it correct? Ah, and before all these, to detect properly the IO expander that actually has that resource in the table (something like gpiod_count() or do we have better approach to answer the question "does this device has a GpioInt() resource?"). -- With Best Regards, Andy Shevchenko
